ORIGINAL E n d o c r i n e
ARTICLE C a r e
Cost-Effectiveness of Molecular Testing for Thyroid Nodules With Atypia of Undetermined Significance Cytology Lawrence Lee, Jacques How, Roger J. Tabah, and Elliot J. Mitmaker Steinberg-Bernstein Centre for Minimally Invasive Surgery and Innovation (L.L.), Department of Surgery (L.L., R.J.T., E.J.M.), and Division of Endocrinology (J.H.), McGill University Health Centre, Montreal, QC H3G 1A4, Canada
Context: Novel molecular diagnostics, such as the gene expression classifier (GEC) and gene mutation panel (GMP) testing, may improve the management for thyroid nodules with atypia of undetermined significance (AUS) cytology. The cost-effectiveness of an approach combining both tests in different practice settings in North America is unknown. Objective: The aim of the study was to determine the cost-effectiveness of two diagnostic molecular tests, singly or in combination, for AUS thyroid nodules. Design and Setting: We constructed a microsimulation model to investigate cost-effectiveness from US (Medicare) and Canadian healthcare system perspectives. Patients: Low-risk patients with AUS thyroid nodules were simulated. Interventions: We examined five management strategies: 1) routine GEC; 2) routine GEC ⫹ selective GMP; 3) routine GMP; 4) routine GMP ⫹ selective GEC; and 5) standard management. Main Outcome Measures: Lifetime costs and quality-adjusted life-years were measured. Results: From the US perspective, the routine GEC ⫹ selective GMP strategy was the dominant strategy. From the Canadian perspective, routine GEC ⫹ selective GMP cost and additional CAN$24 030 per quality-adjusted life-year gained over standard management, and was dominant over the other strategies. Sensitivity analyses reported that the decisions from both perspectives were sensitive to variations in the probability of malignancy in the nodule and the costs of the GEC and GMP. The probability of cost-effectiveness for routine GEC ⫹ selective GMP was low. Conclusions: In the US setting, the most cost-effective strategy was routine GEC ⫹ selective GMP. In the Canadian setting, standard management was most likely to be cost effective. The cost of these molecular diagnostics will need to be reduced to increase their cost-effectiveness for practice settings outside the United States. (J Clin Endocrinol Metab 99: 2674 –2682, 2014)
T
he incidence of thyroid cancer is rapidly increasing in North America (1). Although the reasons for the rising incidence are not fully understood, increased detection of thyroid nodules certainly played a major role, especially as imaging techniques became increasingly more sensitive (2). As more incidental thyroid nodules are identified, the increased burden of further investigation will likely im-
pose a large cost on healthcare systems. Fine-needle aspiration (FNA) biopsy and cytological evaluation are the “gold standard” for the evaluation of thyroid nodules (3); however, up to 30% of FNA samples result in indeterminate cytology (4). Within this category, atypia of undetermined significance (AUS) represents a clinical dilemma because the risk of malignancy (typically ranging from 5 to
ISSN Print 0021-972X ISSN Online 1945-7197 Printed in U.S.A. Copyright © 2014 by the Endocrine Society Received January 26, 2014. Accepted March 18, 2014. First Published Online March 31, 2014
Abbreviations: AUS, atypia of undetermined significance; FNA, fine-needle aspiration; GEC, gene expression classifier; GMP, gene mutation panel; QALY, quality-adjusted lifeyear; RLN, recurrent laryngeal nerve.
2674
jcem.endojournals.org
J Clin Endocrinol Metab, August 2014, 99(8):2674 –2682
doi: 10.1210/jc.2014-1219
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 02 September 2014. at 01:23 For personal use only. No other uses without permission. . All rights reserved.
doi: 10.1210/jc.2014-1219
15%, but it can be as high as 48%) (5) must be balanced with the risk of complications from thyroidectomy. However, new understanding of the molecular pathogenesis of thyroid cancer over the past decade has resulted in novel diagnostic molecular markers that may help identify benign nodules with AUS cytology and consequently avoid unnecessary surgery (6). The two main molecular diagnostic approaches are gene expression from mRNA and screening for genetic mutations. Gene expression testing has been shown to be highly sensitive (90%) but not specific (53%) for malignancy in nodules with AUS cytology (7). On the other hand, gene mutational testing is highly specific (99%) but not sensitive (63%) (8). Therefore, gene expression testing is best used to rule out malignancy in these nodules, whereas gene mutation testing is best used to rule it in. The ideal approach may be a combination of these two approaches. Patients with benign results after gene expression testing may be closely followed without surgery, whereas those with potentially malignant results should undergo further gene mutation testing to help direct the extent of surgery. If mutations are found, then patients may undergo total thyroidectomy as the initial surgery, and if no mutations are found, patients may undergo diagnostic thyroid lobectomy. With this approach, unnecessary and overly aggressive surgery can be avoided. These novel molecular diagnostics are costly, however (6). Although previous economic evaluations have shown positive results for both approaches (9, 10), no analyses have been performed for practice settings outside of the United States, nor has a combination approach been evaluated. The former is especially important, given the discrepancy in US healthcare costs compared to other developed countries, which may significantly affect cost-effectiveness results (11). Therefore, the objective of this study is to determine the cost-effectiveness of a strategy combining gene expression and gene mutation testing for thyroid nodules with AUS from the US and Canadian perspectives.
Subjects and Methods Design An individual patient microsimulation model (1 000 000 patients; 1-y cycle length) was constructed to investigate the lifetime cost-effectiveness of five strategies to manage low-risk patients with thyroid nodules with AUS cytology after two separate FNA cytology samples (Figure 1): 1) routine use of gene expression testing; 2) routine use of a gene expression testing, followed by selective use of gene mutation testing; 3) routine use of gene mutation testing; 4) routine use of gene mutation testing, followed by selective use of gene expression testing; and 5) standard management. Gene expression testing was assumed to have been performed, using a 142 mRNA gene expression classifier (GEC)
jcem.endojournals.org
2675
(7), and gene mutation testing, using a gene mutation panel (GMP) of BRAF, RAS, RET-PTC, and PAX8-PPAR␥ (8). Only patients without major risk factors for thyroid cancer (ie, no history of neck radiation, absence of familial cancer syndromes, nodules ⬍4 cm, node-negative) with AUS on two separate FNA samples according to the Bethesda reporting system for thyroid cytopathology (4) were modeled because these patients represent a clinical dilemma for which molecular diagnostic markers may play an important role in avoiding unnecessary surgery. Similar to a cohort state-transition model, a microsimulation comprises a set of mutually exclusive health states with fixed cycle lengths, each with its set of transition probabilities to other states. However, contrary to the cohort model, which models the cohort as a whole and is limited by the “memory-less” property (called the Markov assumption), a microsimulation model generates each individual subject and follows the individual through the simulation. In this method, subjects can be assigned individual characteristics, and their disease history can be tracked. In the current model, each patient’s age and gender were randomly assigned according to the distributions reported by Alexander et al (7) to account for age- and gender-related probabilities of overall survival. In particular, permanent complications that occurred as a result of surgical interventions (such as recurrent laryngeal nerve [RLN] injury or hypoparathyroidism) were tracked because these affected the lifetime costs and qualityadjusted life-years (QALYs) for each patient. In strategy 1, gene expression testing would be routinely used. Patients with a negative result would then be managed by close clinical observation, defined as per the American Thyroid Association (ATA) guidelines (3). Patients with a false-negative result (ie, malignant pathology falsely diagnosed as benign) would therefore be at risk of delayed diagnosis and treatment. These patients may be at increased risk of recurrence and cancer-specific mortality (12, 13). Patients with a positive test proceed with the standard management, that is, diagnostic thyroid lobectomy, followed by completion thyroidectomy if malignant. In strategy 2, gene expression testing would be routinely performed. If a negative GEC result were obtained, patients would receive close clinical follow-up as per strategy 1. However, if the GEC were positive, then a GMP would be administered. Patients with a positive GMP would undergo total thyroidectomy, whereas patients with a negative GMP would undergo diagnostic lobectomy followed by completion thyroidectomy if malignant. In strategy 3, all patients undergo gene mutation testing. Patients with a positive GMP proceed directly to total thyroidectomy, whereas patients with a negative GMP would undergo standard management as ATA guidelines (3), that is, diagnostic lobectomy followed by completion thyroidectomy if malignant. In strategy 4, gene mutation testing would routinely be performed. A positive GMP would proceed directly to total thyroidectomy as in strategy 3. However, if GMP were negative, then a GEC would be administered, with subsequent management as per strategy 1. Finally, strategy 5 (standard management) consisted of diagnostic lobectomy for all patients with AUS thyroid nodules, followed by completion thyroidectomy if malignant, as per ATA guidelines (3). In all strategies, patients could potentially experience a transient or permanent complication after surgery (unilateral RLN injury after lobectomy; and unilateral or bilateral RLN injury and/or hypoparathyroidism). Transition probabilities for recurrence were obtained from the 10-year recurrence rates for papillary thyroid cancer treated by total thyroidectomy, gathered
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 02 September 2014. at 01:23 For personal use only. No other uses without permission. . All rights reserved.
2676
Lee et al
Molecular Testing for Thyroid Nodules
J Clin Endocrinol Metab, August 2014, 99(8):2674 –2682
Figure 1. Overview of cost-effectiveness model. The squares represent decision nodes, circles represent chance nodes, rectangles represent procedures, and ovals represent health states.
from the National Cancer Database by Bilimoria et al (14). We assumed that recurrences would be locoregional and be treated once by surgery (neck dissection) and ablative radioiodine. Transition probabilities for age- and gender-specific noncancer-re-
lated mortality were obtained from US (15) and Canada (16) life tables, and cancer-related mortality after cancer recurrence was adopted from Rouxel et al (17). Approval from the local institutional review board was not required because no patient in-
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 02 September 2014. at 01:23 For personal use only. No other uses without permission. . All rights reserved.
doi: 10.1210/jc.2014-1219
Table 1.
jcem.endojournals.org
2677
Model Parameters
Parameter Relative risk of mortality with delayed diagnosis Relative risk of recurrence with delayed diagnosis Probability of malignancy in AUS lesions Yearly probability of recurrence after TT and RAI Yearly probability of mortality with recurrent disease Complications after thyroid surgery, % HT–transient RLN injury HT–permanent RLN injury TT–transient RLN injuryb TT–permanent RLN injuryb TT–transient hypoparathyroidism TT–permanent hypoparathyroidism Diagnostic test characteristics, % GEC sensitivityc GEC specificityc GMP sensitivityc GMP specificityc Quality of life weights (“utilities”; 0 ⫽ death, 1 ⫽ perfect health) Unilateral RLN palsy Bilateral RLN palsy Hypoparathyroidism Disease-free after HT Disease-free after TT ⫹ RAI Recurrence Pre-RAI (post-surgery) Post-RAI 0 – 4 wk Post-RAI 4 – 8 wk
Estimate (Range)a
Distributiona
Refs.
2.2 (2.0 –2.4) 2.3 (1.1– 4.9) 14.7 (6.0 – 48.0) 0.8 (0.6 –1.0) 6.9 (5.0 –9.1)
Normal: 2.2, ␦ 0.22 Normal: 2.3, ␦ 1.3 Beta: ␣ 105,  609 Beta: ␣ 1,  124 Beta: ␣ 34,  457
13, 38 12 5 14 17
3.6 (1.9 – 6.2) 2.1 (0.9 – 4.3) 3.6 (2.0 –5.9) 1.0 (0.2–2.6) 27.7 (23.3–32.4) 6.3 (4.1–9.2)
Beta: ␣ 12,  319 Beta: ␣ 7,  324 Beta: ␣ 14,  380 Beta: ␣ 4,  390 Beta: ␣ 109,  285 Beta: ␣ 25,  269
39 39 40 40 40 40
90.3 (74.2–98.0) 53.1 (42.7– 63.2) 62.9 (44.9 –78.5) 98.6 (95.9 –99.7)
Beta: ␣ 28,  3 Beta: ␣ 52,  46 Beta: ␣ 22,  13 Beta: ␣ 209,  3
7 7 8 8
0.63 (0.56 – 0.69) 0.21 (0.18 – 0.23) 0.78 (0.70 – 0.86) 0.99 (0.79 –1.0) 0.95 (0.76 –1.0) 0.54 (0.49 – 0.59) 0.55 (0.49 – 0.60) 0.64 (0.57– 0.70) 0.82 (0.74 – 0.90)
Beta: ␣ 149,  88 Beta: ␣ 318,  1232 Beta: ␣ 88,  25 Beta: ␣ 100,  1 Beta: ␣ 25  1 Beta: ␣ 183  153 Beta: ␣ 180  149 Beta: ␣ 145  82 Beta: ␣ 72  16
27 27 27 25 25 27 26 26 26
Abbreviations: HT, hemithyroidectomy; TT, total thyroidectomy; RAI, radioactive iodine. a
Parameters for distributions were calculated by assuming a SD of 10% of the estimate if no SD was given in the source.
b
Per recurrent laryngeal nerve.
c
AUS only.
formation was accessed, as per the local policy. The model was created and analyzed using TreeAge Pro 2012 (TreeAge Software Inc) and STATA 12 (StataCorp).
Setting The analysis was performed from the US third-party payer (Medicare) and Canadian healthcare system perspectives (ie, societal costs such as productivity losses were not included). Probabilities for outcomes were obtained from specific literature searches (Table 1). If more than one relevant study was identified, then a weighted pooled proportion was calculated using a random-effects model. Future costs and effectiveness were discounted at a 3% rate, but this was varied from 0% to 5% on sensitivity analyses.
Outcomes Costs Costs were expressed as 2013 US$ or CAN$ (1 US$ ⫽ 0.77 CAN$ through purchasing power parity) (18) (Table 2). The costs of the GEC and the GMP were obtained from the respective manufacturers, expressed as US$ and converted into CAN$, because no Canadian-specific pricing is available. The remaining costs of the surgery, adjuvant treatment, and treatment of complications were each derived individually from the US or Canadian perspective. All costs include physician fees. US costs were
obtained from the Healthcare Cost and Utilization Project from the Agency of Healthcare Research and Quality (19) using average 2011 Medicare costs for the appropriate ICD-9 procedure codes, and from the 2013 Medicare physician and clinical laboratory fee schedules (2013 conversion factor ⫽ US$25.00) (20). Canadian (province of Quebec) costs were obtained from the Canadian Institute for Health Information case-mix groups (21), which are similar to diagnosis-related groups. These costs represent the mean overall cost of the procedure, including the operation and subsequent hospitalization. Other costs and physician fees were obtained from Quebec provincial cost manuals (22, 23). Drug costs were based on US (24) and Quebec provincial (23) wholesale costs.
Effectiveness QALYs were used as the main effectiveness measure. QALYs are calculated by multiplying the utility of a health state by the time spent in that health state. Utilities (also known as quality of life weights) are valued from a scale of 0 (death) to 1 (perfect health). Utilities for each of the health states in our model were taken from three different studies (Table 1) (25–27). Costs and effectiveness measures were incorporated into a single summary measure as the incremental cost-effectiveness ratios.
Sensitivity analysis Deterministic and probabilistic sensitivity analyses were performed to account for parameter uncertainty. Deterministic sen-
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 02 September 2014. at 01:23 For personal use only. No other uses without permission. . All rights reserved.
2678
Lee et al
Molecular Testing for Thyroid Nodules
J Clin Endocrinol Metab, August 2014, 99(8):2674 –2682
Table 2. Costs of Parameters Included in the Model From the US Third-Party Payer and Canadian Healthcare System Perspectives United States, 2013 $US
GEC GMP Thyroid lobectomy Total thyroidectomy Completion thyroidectomy Neck dissection Radioactive iodine ablation (whole body I131 scan, I131 ablation, 1 inpatient day) Yearly cost of surveillance (TSH, thyroglobulin, anti-thyroglobulin antibodies, neck ultrasound, and physical examination) Yearly cost of follow-up of benign nodules (neck ultrasound, physical examination) Yearly cost of TSH suppression (assuming daily dose of levothyroxine 100 g) Yearly cost of treatment of hypoparathyroidism (assuming daily dose of calcitriol 25 g and calcium/vitamin D 1000 mg/800 U)
Canada, 2013 $CAN
Estimatea (rangeb)c
Source or Ref.
Estimatea (rangeb)c
Source or Ref.
3500 (1750 –7000) 850 (425–1700) 9042 (2521–18 084) 10 155 (5078 –20 310) 10 364 (5182–20 728) 16 729 (8365–33 458) 451 (226 –902)
Manufacturer Manufacturer 19, 20 19, 20 19, 20 19, 20 20
4340 (2170 – 8680) 1054 (527–2108) 4205 (2103– 8410) 4483 (2242– 8966) 4359 (2180 – 8718) 6120 (3060 –12 240) 1580 (790 –3160)
Manufacturer Manufacturer 21, 22 21, 22 21, 22 21, 22 22, 23
309 (155– 618)
20
220 (110 – 440)
22, 23
242 (121– 484)
20
136 (68 –272)
22, 23
147 (74 –294)
24
22 (11– 44)
23
242 (121– 484)
24
453 (227–906)
23
a
All costs include physicians’ fees.
b
Ranges were calculated using 50% of the base estimate for the lower limit and 200% for the upper limit.
␥-Distributions were fitted around each cost estimate using ␣ ⫽ 1 and  ⫽ base case cost as parameters, except for the costs of the GEC and the GMP, which were not subject to probabilistic uncertainty. c
sitivity analysis was performed by varying one or more variables at a time across a specified range of values. Ranges were derived from the literature (directly or from pooled analyses). Microsimulation models inherently perform probabilistic sensitivity analysis because parameters for each simulated patient are randomly drawn from the distributions around each variable. The appropriate distributions were fitted around each variable (normal distribution for relative risks,  distribution for probabilities and utilities, and ␥ distribution for costs), which were derived from the literature. The 95% confidence intervals around the point estimates were obtained from the 2.5th and 97.5th percentiles. Uncertainty around cost-effectiveness results was represented through cost-effectiveness acceptability curves, which represent the proportion of times (out of the total number of simulated patients) that each option is optimal for a given cost/ QALY threshold.
Results The mean age of the simulated patients was 54.4 years (SD, 21.3), and 19% were male. Results of the cost-effectiveness analyses from the US and Canadian perspectives are reported in Table 3. From the US perspective, routine use of GEC with selective use of GMP resulted in the lowest cost and the most QALYs (ie, dominant) compared to the other four strategies. From the Canadian perspective, this strategy was dominant over the routine use of GEC
only, but it resulted in an additional CAN$24 030 per QALY gained when compared to standard management. From both perspectives, the routine use of the GEC-only strategy was the most costly. As well, the routine use of GMP and the routine use of GMP with selective use of GEC strategies were dominated (ie, more costly and less effective) from both perspectives. There was no difference in life expectancy across strategies from either perspective (Table 3). Similarly, there was minimal difference in QALYs between strategies. The cost-effectiveness results were most sensitive to variations in the cost of the GEC and GMP (Figure 2, A and B). Sensitivity analyses demonstrated that in order for the routine GEC ⫹ selective GMP strategy to be less costly than standard management, the GEC needed to cost less than US$5036 and the GMP less than US$1609 from the US perspective. From the Canadian perspective, this strategy was the least costly only if the GEC cost less than CAN$2215 and the GMP less than CAN$721. From both perspectives, the routine GMP and the routine GMP ⫹ selective GEC strategies were never the least costly option across the range of values. From the US perspective, the routine GEC ⫹ selective GMP strategy would be the least costly only if the cost of a thyroid lobectomy exceeded US$6686. Other variables tested had minimal impact (data not shown).
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 02 September 2014. at 01:23 For personal use only. No other uses without permission. . All rights reserved.
doi: 10.1210/jc.2014-1219
jcem.endojournals.org
2679
Table 3. Cost-Effectiveness Results From Both the US Third-Party Payer and Canadian Healthcare System Perspectives
US third-party payer perspective, 2013 US$ Standard management Routine GEC Routine GEC with selective GMP Routine GMP Routine GMP with selective GEC Canadian healthcare system perspective, 2013 CAN$ Standard management Routine GEC Routine GEC with selective GMP Routine GMP Routine GMP with selective GEC
Mean Life Expectancy, y (95% CI)
Mean Cost (95% CIa)
27.64b (27.24, 28.04)
22 385 (22 178, 22 593)
27.41 (27.02, 27.82) 27.53 (27.13, 27.93)
22 586 (22 331, 22 846) 20 506 (20 287, 20 732)
⫹201 (56, 348) ⫺2080 (⫺2285, ⫺1875)
27.65 (27.24, 28.05) 27.45 (27.05, 27.84)
21 588 (21 396, 21 776) 21 267 (21 040, 21 491)
28.52e (28.11, 28.93) 28.57 (28.15, 28.98) 28.78 (28.37, 29.20) 28.28 (27.87, 28.69) 28.44 (28.03, 28.86)
Incremental Cost (95% CI)
Mean QALYs (95% CI)
Incremental QALYs (95% CI)
ICER, $/QALY
17.23 (17.06, 17.39) 17.28 (17.12, 17.45)
⫹0.12 (0.01, 0.23) ⫹0.03 (⫺0.05, 0.12)
16 750 Dominantc,d
⫹1082 (780, 1383) ⫺320 (⫺562, ⫺78)
17.12 (16.97, 17.29) 17.09 (16.93, 17.25)
⫺0.26 (⫺0.36, ⫺0.15) ⫹0.10 (⫺0.01, 0.22)
Dominated Dominated
11 050 (10 942, 11 157) 13 814 (13 685, 13 941) 13 453 (13 335, 13 573)
⫹2764 (2692, 2835) ⫺360 (⫺464, ⫺257)
17.18 (17.02, 17.35) 17.25 (17.09, 17.41) 17.28 (17.12, 17.45)
⫹0.07 (⫺0.04, 0.19) ⫹0.05 (⫺0.03, 0.14)
39485 24 030,c Dominantd
11 851 (11 738, 11 963) 13 640 (13 519, 13 756)
⫺1602 (⫺1786, ⫺1418) ⫹1788 (1644, 1933)
17.02 (16.86, 17.19) 17.12 (16.96, 17.29)
⫺0.15 (⫺0.26, ⫺0.04) ⫺0.04 (⫺0.15, 0.07)
Dominated Dominated
17.11 (16.94, 17.27)
Abbreviations: CI, confidence interval; ICER, incremental cost-effectiveness ratio; Dominant, less costly and more effective; Dominated, more costly and/or less effective. a
CIs around mean costs derived from bootstrapped estimates. Other CIs are normal distribution-based.
b
P value ⫽ .552 for the differences in life expectancy between strategies from the US perspective.
c
Compared to standard management.
d
Compared to routine GEC strategy.
e
P value ⫽ .895 for the differences in life expectancy between strategies from the Canadian perspective.
The results were also sensitive to variations in the probability of malignancy from both perspectives. The interaction between the costs of the GEC and the probability of malignancy demonstrated that the routine GEC strategy was the least costly strategy only in cases when the cost of the GEC and probability of malignancy were both low, although the
threshold values were different for each perspective (Figure 2, C and D). Again, the routine GMP and the routine GMP ⫹ selective GEC strategies were never the least costly option across the range of values. Cost-effectiveness acceptability curves are shown in Figure 3, A and B. From the US perspective, the routine GEC ⫹ selective GMP strategy had the highest probability of cost-effectiveness across all cost/QALY thresholds, whereas from the Canadian perspective, standard management was most likely to be cost effective. However, there was significant uncertainty around these results from both perspectives because none of the strategies had a probability of cost-effectiveness higher than 35% at any cost/ QALY threshold.
Discussion Figure 2. Relationship between variations in cost of GEC and GMP on incremental strategy costs from the US third-party payer (A) and Canadian healthcare system (B) perspectives; and between variations in the cost of the GEC and risk of malignancy in nodule from the US thirdparty payer (C) and Canadian healthcare system (D) perspectives.
Thyroid nodules with AUS represent an important clinical dilemma. Novel molecular diagnostic markers, such as gene expression and gene muta-
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 02 September 2014. at 01:23 For personal use only. No other uses without permission. . All rights reserved.
2680
Lee et al
Molecular Testing for Thyroid Nodules
J Clin Endocrinol Metab, August 2014, 99(8):2674 –2682
Figure 3. Cost-effectiveness acceptability curves from the US third-party payer (A) and Canadian healthcare system (B) perspectives.
tion testing, may help identify nodules that are more likely to be benign or malignant and help avoid unnecessary surgery and the risks associated with thyroidectomy (28). However, these novel markers are costly, and their costeffectiveness has yet to be fully evaluated. In the present study, a strategy combining routine GEC and selective GMP use for AUS thyroid nodules was most likely to be cost effective in the US setting, whereas standard management was the strategy with the highest probability of costeffectiveness from a Canadian perspective. Although the cost-effectiveness results from both perspectives appear favorable for the strategy combining routine GEC ⫹ selective GMP, there was considerable uncertainty around the estimates, as demonstrated in the cost-effectiveness acceptability curves (Figure 3, A and B). This was also reflected in deterministic sensitivity analyses, which reported important changes in the interpretation of the cost-effectiveness results based on variations in the cost of the GEC, diagnostic thyroid lobectomy, and the probability of malignancy. There were also important differences between the US and Canadian perspectives. In the United States, there were cost savings because of the high cost of thyroid lobectomy relative to the cost of the molecular markers, whereas in Canada, the low cost of surgery relative to the molecular markers resulted in higher costs associated with their use. Favorable cost-effectiveness results are obtained with current US pricing, but the price of both tests will have to be lower in Canada before molecular marker diagnostics becomes cost saving. Also, the addition of gene mutation testing to routine gene expression testing becomes less costly overall as the probability of malignancy becomes higher because more patients with a positive GEC result would receive singlestage total thyroidectomy, as long as the cost of the GEC remains (relatively) low. Differences in the cost of the competing strategies were more important compared to differences in their effective-
ness because there were minimal to nonexistent differences in life expectancy or QALYs between the strategies (slight differences between the US and Canadian settings were due to differences in age-specific mortality estimates between the two countries). This can be explained by the fact that the risks of surgical complications that would have an important negative impact on quality of life, such as RLN, and thyroid cancer-related mortality were low (29, 30). Thus, the cost-effectiveness results were mostly dependent on the minimization of unnecessary surgery. The use of gene mutation testing in nodules with AUS cytology is controversial. Previous studies have reported that testing for gene mutations was highly specific (with high positive predictive value) for malignancy but had low sensitivity. Therefore, gene mutation testing is ideal to determine the extent of surgery in patients with a high suspicion of malignancy because those patients who test positive for a gene mutation can avoid a two-stage operation (diagnostic thyroid lobectomy, followed by completion thyroidectomy if malignant) and proceed directly to total thyroidectomy. However, the diagnostic characteristics of gene mutation testing (ie, low sensitivity and negative predictive value) do not allow for the accurate identification of nodules with relatively low probability of malignancy (6). Furthermore, there are data reporting that a high proportion of follicular variant of papillary thyroid cancers do not express the BRAF mutation and therefore may not register positive on GMP testing (31, 32). Therefore, patients with a negative GMP result should still undergo diagnostic lobectomy, given the increased risk of false negatives. The selective use of a GEC after a negative GMP result may reduce the number of false negatives and unnecessary surgery, but these two strategies (routine GMP only and routine GMP ⫹ selective GEC) were not costeffective from either perspective. There have been previous studies that investigated the cost-effectiveness of gene expression and gene mutation
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 02 September 2014. at 01:23 For personal use only. No other uses without permission. . All rights reserved.
doi: 10.1210/jc.2014-1219
testing for thyroid nodules with indeterminate cytology. Li et al (10) performed a model-based economic evaluation of routine GEC use compared to standard practice in the United States over a 5-year time horizon and reported cost savings with a small increase in QALYs. Yip et al (9) also used a model-based approach to perform a cost analysis comparing routine GMP use to standard practice in the United States and reported favorable results. No studies have performed an economic evaluation alongside a clinical trial or using individual-patient data. There are some important differences between these studies and our results. In the present study, routine GEC was associated with increased costs, whereas both previous studies reported cost savings (10). This discrepancy may be explained in part by differences in model parameters, model type and structure, and time horizon. For example, Li et al (10) modeled cost-effectiveness results over a 5-year time horizon, which may not be long enough for a slowly progressive disease such as thyroid cancer. Furthermore, these two studies used less sophisticated models (with more constraints) than the present study. Although all model-based evaluations are simplified representations of reality, a less constrained model that more accurately mimics the natural history of a disease, such as microsimulation, is more likely to provide realistic results (33). Lastly, the present study evaluated two additional combination strategies of routine GEC ⫹ selective GMP and routine GMP ⫹ selective GEC, and performed the economic evaluation from multiple perspectives. There are also some preliminary data to suggest that papillary microcarcinomas (less than 1 cm) diagnosed on FNA can be observed, and resected only if they progress, with similar outcomes as outright resection (34, 35). Although this approach may certainly affect the cost-effectiveness of the management strategies that were evaluated in the present study, it is not directly germane to our objectives and data obtained because our target population has not yet been diagnosed with thyroid cancer and is still undergoing diagnostic investigations. This study has limitations. As with any model-based economic evaluation, the inputs for model parameters are based on data from other studies and are subject to the limitations of the source studies themselves. The diagnostic performance characteristics of the GEC and the GMP were based on two studies (7, 8). It is possible that the diagnostic performance characteristics of either test may change as more validation studies are performed. Other model inputs were obtained largely from single-center observational studies. Also, the results of this study may not be generalizable across all practice settings because our results demonstrate that the cost-effectiveness of the competing strategies was highly sensitive to variations in the
jcem.endojournals.org
2681
cost of important parameters. This is especially important given that previous studies have shown wide variations in the cost of care depending on the Medicare region (36). There were also certain assumptions that were made regarding treatment options that remain controversial, such as routine radioactive iodine remnant ablation and the choice of total thyroidectomy for definitive treatment (37). In these cases, we modeled management recommendations based on ATA guidelines (3). In summary, the results of this economic evaluation suggest that a strategy combining routine gene expression testing with selective gene mutation testing may be cost effective in the management of patients with AUS thyroid nodules. However, this decision was subject to considerable uncertainty because the cost-effectiveness of this strategy was different in the United States than in Canada and was highly sensitive to important model parameters, such as the cost of the GEC, thyroid lobectomy, and the probability of malignancy in the nodule. In order for molecular diagnostic markers to be more cost effective in a setting outside the United States, the cost of these tests needs to be revised.
Acknowledgments Address all correspondence and requests for reprints to: Lawrence Lee, MD, MSc, Steinberg-Bernstein Centre for Minimally Invasive Surgery and Innovation, Department of Surgery, McGill University Health Centre, 1650 Cedar Avenue, E19-125, Montreal, QC H3G 1A4, Canada. E-mail: [email protected]. Disclosure Summary: The authors have nothing to declare.
References 1. Davies L, Welch HG. Increasing incidence of thyroid cancer in the United States, 1973–2002. JAMA. 2006;295:2164 –2167. 2. Grodski S, Brown T, Sidhu S, et al. Increasing incidence of thyroid cancer is due to increased pathologic detection. Surgery. 2008;144: 1038 –1043. 3. Cooper DS, Doherty GM, Haugen BR, et al. Revised American Thyroid Association management guidelines for patients with thyroid nodules and differentiated thyroid cancer. Thyroid. 2009;19:1167– 1214. 4. Cibas ES, Ali SZ. The Bethesda system for reporting thyroid cytopathology. Am J Clin Pathol. 2009;132:658 – 665. 5. Wang CC, Friedman L, Kennedy GC, et al. A large multicenter correlation study of thyroid nodule cytopathology and histopathology. Thyroid. 2011;21:243–251. 6. Xing M, Haugen BR, Schlumberger M. Progress in molecular-based management of differentiated thyroid cancer. Lancet. 2013;381: 1058 –1069. 7. Alexander EK, Kennedy GC, Baloch ZW, et al. Preoperative diagnosis of benign thyroid nodules with indeterminate cytology. N Engl J Med. 2012;367:705–715. 8. Nikiforov YE, Ohori NP, Hodak SP, et al. Impact of mutational
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 02 September 2014. at 01:23 For personal use only. No other uses without permission. . All rights reserved.
2682
9.
10.
11. 12.
13.
14.
15.
16.
17.
18.
19.
20.
21. 22.
23.
24. 25.
Lee et al
Molecular Testing for Thyroid Nodules
testing on the diagnosis and management of patients with cytologically indeterminate thyroid nodules: a prospective analysis of 1056 FNA samples. J Clin Endocrinol Metab. 2011;96:3390 –3397. Yip L, Farris C, Kabaker AS, et al. Cost impact of molecular testing for indeterminate thyroid nodule fine-needle aspiration biopsies. J Clin Endocrinol Metab. 2012;97:1905–1912. Li H, Robinson KA, Anton B, Saldanha IJ, Ladenson PW. Costeffectiveness of a novel molecular test for cytologically indeterminate thyroid nodules. J Clin Endocrinol Metab. 2011;96:E1719 – E1726. Bodenheimer T. High and rising health care costs. Part 1: seeking an explanation. Ann Intern Med. 2005;142:847– 854. Yeh MW, Demircan O, Ituarte P, Clark OH. False-negative fineneedle aspiration cytology results delay treatment and adversely affect outcome in patients with thyroid carcinoma. Thyroid. 2004; 14:207–215. Najafzadeh M, Marra CA, Lynd LD, Wiseman SM. Cost-effectiveness of using a molecular diagnostic test to improve preoperative diagnosis of thyroid cancer. Value Health. 2012;15:1005–1013. Bilimoria KY, Bentrem DJ, Ko CY, et al. Extent of surgery affects survival for papillary thyroid cancer. Ann Surg. 2007;246:375–381; discussion 381–384. Arias E. United States life tables, 2008. National Vital Statistics Reports. Vol 61, no. 3. http://www.cdc.gov/nchs/data/nvsr/nvsr61/ nvsr61_03.pdf. Accessed July 12, 2013. Life Tables, Canada, Provinces and Territories 2007 to 2009. http://www.statcan.gc.ca/pub/84-537-x/84-537-x2013003-eng. htm. Accessed July 1, 2013. Rouxel A, Hejblum G, Bernier MO, et al. Prognostic factors associated with the survival of patients developing loco-regional recurrences of differentiated thyroid carcinomas. J Clin Endocrinol Metab. 2004;89:5362–5368. Organization for Economic Cooperation and Development. 2008 PPP Benchmark results. http://stats.oecd.org/Index.aspx?DataSetCode⫽CPL. Accessed January 6, 2013. Agency for Healthcare Research and Quality. Healthcare Cost and Utilization Project (HCUP.net). http://hcupnet.ahrq.gov/. Accessed October 15, 2013. Centers for Medicare & Medicaid Services. Physician fee schedule. http://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/ PhysicianFeeSched/. Accessed October 22, 2013. Canadian Institute for Health Information. Case mix groups (CMG⫹). http://www.cihi.ca/casemix. Accessed March 30, 2013. Régie de l’assurance maladie du Québec. Manuel de facturation. http:// www.ramq.gouv.qc.ca/fr/professionnels/medecins-specialistes/manuels/ Pages/facturation.aspx. Accessed May 1, 2013. Ministère de la Santé et des Services sociaux. Normes et pratiques de gestion pour les centres hospitaliers. http://msssa4.msss.gouv.qc.ca/ fr/document/d26ngest.nsf/lcaCH?OpenView. Accessed October 22, 2013. Medi-Span. http://www.medispan.com/. Accessed October 30, 2013. Esnaola NF, Cantor SB, Sherman SI, Lee JE, Evans DB. Optimal
J Clin Endocrinol Metab, August 2014, 99(8):2674 –2682
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
treatment strategy in patients with papillary thyroid cancer: a decision analysis. Surgery. 2001;130:921–930. Mernagh P, Suebwongpat A, Silverberg J, Weston A. Cost-effectiveness of using recombinant human thyroid-stimulating hormone before radioiodine ablation for thyroid cancer: the Canadian perspective. Value Health. 2010;13:180 –187. Kebebew E, Duh QY, Clark OH. Total thyroidectomy or thyroid lobectomy in patients with low-risk differentiated thyroid cancer: surgical decision analysis of a controversy using a mathematical model. World J Surg. 2000;24:1295–1302. Alexander EK, Schorr M, Klopper J, et al. Multicenter clinical experience with the Afirma gene expression classifier. J Clin Endocrinol Metab. 2014;99:119 –125. Rosato L, Avenia N, Bernante P, et al. Complications of thyroid surgery: analysis of a multicentric study on 14,934 patients operated on in Italy over 5 years. World J Surg. 2004;28:271–276. Hundahl SA, Fleming ID, Fremgen AM, Menck HR. A National Cancer Data Base report on 53,856 cases of thyroid carcinoma treated in the U.S., 1985–1995. Cancer. 1998;83:2638 –2648. Lim JY, Hong SW, Lee YS, et al. Clinicopathologic implications of the BRAF(V600E) mutation in papillary thyroid cancer: a subgroup analysis of 3130 cases in a single center. Thyroid. 2013;23:1423– 1430. Trovisco V, Soares P, Preto A, et al. Type and prevalence of BRAF mutations are closely associated with papillary thyroid carcinoma histotype and patients’ age but not with tumour aggressiveness. Virchows Arch. 2005;446:589 –595. Weinstein MC. Recent developments in decision-analytic modelling for economic evaluation. Pharmacoeconomics. 2006;24:1043– 1053. Sugitani I, Toda K, Yamada K, Yamamoto N, Ikenaga M, Fujimoto Y. Three distinctly different kinds of papillary thyroid microcarcinoma should be recognized: our treatment strategies and outcomes. World J Surg. 2010;34:1222–1231. Ito Y, Miyauchi A, Kihara M, Higashiyama T, Kobayashi K, Miya A. Patient age is significantly related to the progression of papillary microcarcinoma of the thyroid under observation. Thyroid. 2014; 24:27–34. Fisher ES, Wennberg DE, Stukel TA, Gottlieb DJ, Lucas FL, Pinder EL. The implications of regional variations in Medicare spending. Part 1: the content, quality, and accessibility of care. Ann Intern Med. 2003;138:273–287. McLeod DS, Sawka AM, Cooper DS. Controversies in primary treatment of low-risk papillary thyroid cancer. Lancet. 2013;381: 1046 –1057. Mazzaferri EL, Jhiang SM. Long-term impact of initial surgical and medical therapy on papillary and follicular thyroid cancer. Am J Med. 1994;97:418 – 428. Landerholm K, Wasner AM, Järhult J. Incidence and risk factors for injuries to the recurrent laryngeal nerve during neck surgery in the moderate-volume setting [published online January 9, 2014]. Langenbecks Arch Surg. doi:10.1007/s00423-013-1154-6. Giordano D, Valcavi R, Thompson GB, et al. Complications of central neck dissection in patients with papillary thyroid carcinoma: results of a study on 1087 patients and review of the literature. Thyroid. 2012;22:911–917.
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 02 September 2014. at 01:23 For personal use only. No other uses without permission. . All rights reserved.
ARTICLE C a r e
Cost-Effectiveness of Molecular Testing for Thyroid Nodules With Atypia of Undetermined Significance Cytology Lawrence Lee, Jacques How, Roger J. Tabah, and Elliot J. Mitmaker Steinberg-Bernstein Centre for Minimally Invasive Surgery and Innovation (L.L.), Department of Surgery (L.L., R.J.T., E.J.M.), and Division of Endocrinology (J.H.), McGill University Health Centre, Montreal, QC H3G 1A4, Canada
Context: Novel molecular diagnostics, such as the gene expression classifier (GEC) and gene mutation panel (GMP) testing, may improve the management for thyroid nodules with atypia of undetermined significance (AUS) cytology. The cost-effectiveness of an approach combining both tests in different practice settings in North America is unknown. Objective: The aim of the study was to determine the cost-effectiveness of two diagnostic molecular tests, singly or in combination, for AUS thyroid nodules. Design and Setting: We constructed a microsimulation model to investigate cost-effectiveness from US (Medicare) and Canadian healthcare system perspectives. Patients: Low-risk patients with AUS thyroid nodules were simulated. Interventions: We examined five management strategies: 1) routine GEC; 2) routine GEC ⫹ selective GMP; 3) routine GMP; 4) routine GMP ⫹ selective GEC; and 5) standard management. Main Outcome Measures: Lifetime costs and quality-adjusted life-years were measured. Results: From the US perspective, the routine GEC ⫹ selective GMP strategy was the dominant strategy. From the Canadian perspective, routine GEC ⫹ selective GMP cost and additional CAN$24 030 per quality-adjusted life-year gained over standard management, and was dominant over the other strategies. Sensitivity analyses reported that the decisions from both perspectives were sensitive to variations in the probability of malignancy in the nodule and the costs of the GEC and GMP. The probability of cost-effectiveness for routine GEC ⫹ selective GMP was low. Conclusions: In the US setting, the most cost-effective strategy was routine GEC ⫹ selective GMP. In the Canadian setting, standard management was most likely to be cost effective. The cost of these molecular diagnostics will need to be reduced to increase their cost-effectiveness for practice settings outside the United States. (J Clin Endocrinol Metab 99: 2674 –2682, 2014)
T
he incidence of thyroid cancer is rapidly increasing in North America (1). Although the reasons for the rising incidence are not fully understood, increased detection of thyroid nodules certainly played a major role, especially as imaging techniques became increasingly more sensitive (2). As more incidental thyroid nodules are identified, the increased burden of further investigation will likely im-
pose a large cost on healthcare systems. Fine-needle aspiration (FNA) biopsy and cytological evaluation are the “gold standard” for the evaluation of thyroid nodules (3); however, up to 30% of FNA samples result in indeterminate cytology (4). Within this category, atypia of undetermined significance (AUS) represents a clinical dilemma because the risk of malignancy (typically ranging from 5 to
ISSN Print 0021-972X ISSN Online 1945-7197 Printed in U.S.A. Copyright © 2014 by the Endocrine Society Received January 26, 2014. Accepted March 18, 2014. First Published Online March 31, 2014
Abbreviations: AUS, atypia of undetermined significance; FNA, fine-needle aspiration; GEC, gene expression classifier; GMP, gene mutation panel; QALY, quality-adjusted lifeyear; RLN, recurrent laryngeal nerve.
2674
jcem.endojournals.org
J Clin Endocrinol Metab, August 2014, 99(8):2674 –2682
doi: 10.1210/jc.2014-1219
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 02 September 2014. at 01:23 For personal use only. No other uses without permission. . All rights reserved.
doi: 10.1210/jc.2014-1219
15%, but it can be as high as 48%) (5) must be balanced with the risk of complications from thyroidectomy. However, new understanding of the molecular pathogenesis of thyroid cancer over the past decade has resulted in novel diagnostic molecular markers that may help identify benign nodules with AUS cytology and consequently avoid unnecessary surgery (6). The two main molecular diagnostic approaches are gene expression from mRNA and screening for genetic mutations. Gene expression testing has been shown to be highly sensitive (90%) but not specific (53%) for malignancy in nodules with AUS cytology (7). On the other hand, gene mutational testing is highly specific (99%) but not sensitive (63%) (8). Therefore, gene expression testing is best used to rule out malignancy in these nodules, whereas gene mutation testing is best used to rule it in. The ideal approach may be a combination of these two approaches. Patients with benign results after gene expression testing may be closely followed without surgery, whereas those with potentially malignant results should undergo further gene mutation testing to help direct the extent of surgery. If mutations are found, then patients may undergo total thyroidectomy as the initial surgery, and if no mutations are found, patients may undergo diagnostic thyroid lobectomy. With this approach, unnecessary and overly aggressive surgery can be avoided. These novel molecular diagnostics are costly, however (6). Although previous economic evaluations have shown positive results for both approaches (9, 10), no analyses have been performed for practice settings outside of the United States, nor has a combination approach been evaluated. The former is especially important, given the discrepancy in US healthcare costs compared to other developed countries, which may significantly affect cost-effectiveness results (11). Therefore, the objective of this study is to determine the cost-effectiveness of a strategy combining gene expression and gene mutation testing for thyroid nodules with AUS from the US and Canadian perspectives.
Subjects and Methods Design An individual patient microsimulation model (1 000 000 patients; 1-y cycle length) was constructed to investigate the lifetime cost-effectiveness of five strategies to manage low-risk patients with thyroid nodules with AUS cytology after two separate FNA cytology samples (Figure 1): 1) routine use of gene expression testing; 2) routine use of a gene expression testing, followed by selective use of gene mutation testing; 3) routine use of gene mutation testing; 4) routine use of gene mutation testing, followed by selective use of gene expression testing; and 5) standard management. Gene expression testing was assumed to have been performed, using a 142 mRNA gene expression classifier (GEC)
jcem.endojournals.org
2675
(7), and gene mutation testing, using a gene mutation panel (GMP) of BRAF, RAS, RET-PTC, and PAX8-PPAR␥ (8). Only patients without major risk factors for thyroid cancer (ie, no history of neck radiation, absence of familial cancer syndromes, nodules ⬍4 cm, node-negative) with AUS on two separate FNA samples according to the Bethesda reporting system for thyroid cytopathology (4) were modeled because these patients represent a clinical dilemma for which molecular diagnostic markers may play an important role in avoiding unnecessary surgery. Similar to a cohort state-transition model, a microsimulation comprises a set of mutually exclusive health states with fixed cycle lengths, each with its set of transition probabilities to other states. However, contrary to the cohort model, which models the cohort as a whole and is limited by the “memory-less” property (called the Markov assumption), a microsimulation model generates each individual subject and follows the individual through the simulation. In this method, subjects can be assigned individual characteristics, and their disease history can be tracked. In the current model, each patient’s age and gender were randomly assigned according to the distributions reported by Alexander et al (7) to account for age- and gender-related probabilities of overall survival. In particular, permanent complications that occurred as a result of surgical interventions (such as recurrent laryngeal nerve [RLN] injury or hypoparathyroidism) were tracked because these affected the lifetime costs and qualityadjusted life-years (QALYs) for each patient. In strategy 1, gene expression testing would be routinely used. Patients with a negative result would then be managed by close clinical observation, defined as per the American Thyroid Association (ATA) guidelines (3). Patients with a false-negative result (ie, malignant pathology falsely diagnosed as benign) would therefore be at risk of delayed diagnosis and treatment. These patients may be at increased risk of recurrence and cancer-specific mortality (12, 13). Patients with a positive test proceed with the standard management, that is, diagnostic thyroid lobectomy, followed by completion thyroidectomy if malignant. In strategy 2, gene expression testing would be routinely performed. If a negative GEC result were obtained, patients would receive close clinical follow-up as per strategy 1. However, if the GEC were positive, then a GMP would be administered. Patients with a positive GMP would undergo total thyroidectomy, whereas patients with a negative GMP would undergo diagnostic lobectomy followed by completion thyroidectomy if malignant. In strategy 3, all patients undergo gene mutation testing. Patients with a positive GMP proceed directly to total thyroidectomy, whereas patients with a negative GMP would undergo standard management as ATA guidelines (3), that is, diagnostic lobectomy followed by completion thyroidectomy if malignant. In strategy 4, gene mutation testing would routinely be performed. A positive GMP would proceed directly to total thyroidectomy as in strategy 3. However, if GMP were negative, then a GEC would be administered, with subsequent management as per strategy 1. Finally, strategy 5 (standard management) consisted of diagnostic lobectomy for all patients with AUS thyroid nodules, followed by completion thyroidectomy if malignant, as per ATA guidelines (3). In all strategies, patients could potentially experience a transient or permanent complication after surgery (unilateral RLN injury after lobectomy; and unilateral or bilateral RLN injury and/or hypoparathyroidism). Transition probabilities for recurrence were obtained from the 10-year recurrence rates for papillary thyroid cancer treated by total thyroidectomy, gathered
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 02 September 2014. at 01:23 For personal use only. No other uses without permission. . All rights reserved.
2676
Lee et al
Molecular Testing for Thyroid Nodules
J Clin Endocrinol Metab, August 2014, 99(8):2674 –2682
Figure 1. Overview of cost-effectiveness model. The squares represent decision nodes, circles represent chance nodes, rectangles represent procedures, and ovals represent health states.
from the National Cancer Database by Bilimoria et al (14). We assumed that recurrences would be locoregional and be treated once by surgery (neck dissection) and ablative radioiodine. Transition probabilities for age- and gender-specific noncancer-re-
lated mortality were obtained from US (15) and Canada (16) life tables, and cancer-related mortality after cancer recurrence was adopted from Rouxel et al (17). Approval from the local institutional review board was not required because no patient in-
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 02 September 2014. at 01:23 For personal use only. No other uses without permission. . All rights reserved.
doi: 10.1210/jc.2014-1219
Table 1.
jcem.endojournals.org
2677
Model Parameters
Parameter Relative risk of mortality with delayed diagnosis Relative risk of recurrence with delayed diagnosis Probability of malignancy in AUS lesions Yearly probability of recurrence after TT and RAI Yearly probability of mortality with recurrent disease Complications after thyroid surgery, % HT–transient RLN injury HT–permanent RLN injury TT–transient RLN injuryb TT–permanent RLN injuryb TT–transient hypoparathyroidism TT–permanent hypoparathyroidism Diagnostic test characteristics, % GEC sensitivityc GEC specificityc GMP sensitivityc GMP specificityc Quality of life weights (“utilities”; 0 ⫽ death, 1 ⫽ perfect health) Unilateral RLN palsy Bilateral RLN palsy Hypoparathyroidism Disease-free after HT Disease-free after TT ⫹ RAI Recurrence Pre-RAI (post-surgery) Post-RAI 0 – 4 wk Post-RAI 4 – 8 wk
Estimate (Range)a
Distributiona
Refs.
2.2 (2.0 –2.4) 2.3 (1.1– 4.9) 14.7 (6.0 – 48.0) 0.8 (0.6 –1.0) 6.9 (5.0 –9.1)
Normal: 2.2, ␦ 0.22 Normal: 2.3, ␦ 1.3 Beta: ␣ 105,  609 Beta: ␣ 1,  124 Beta: ␣ 34,  457
13, 38 12 5 14 17
3.6 (1.9 – 6.2) 2.1 (0.9 – 4.3) 3.6 (2.0 –5.9) 1.0 (0.2–2.6) 27.7 (23.3–32.4) 6.3 (4.1–9.2)
Beta: ␣ 12,  319 Beta: ␣ 7,  324 Beta: ␣ 14,  380 Beta: ␣ 4,  390 Beta: ␣ 109,  285 Beta: ␣ 25,  269
39 39 40 40 40 40
90.3 (74.2–98.0) 53.1 (42.7– 63.2) 62.9 (44.9 –78.5) 98.6 (95.9 –99.7)
Beta: ␣ 28,  3 Beta: ␣ 52,  46 Beta: ␣ 22,  13 Beta: ␣ 209,  3
7 7 8 8
0.63 (0.56 – 0.69) 0.21 (0.18 – 0.23) 0.78 (0.70 – 0.86) 0.99 (0.79 –1.0) 0.95 (0.76 –1.0) 0.54 (0.49 – 0.59) 0.55 (0.49 – 0.60) 0.64 (0.57– 0.70) 0.82 (0.74 – 0.90)
Beta: ␣ 149,  88 Beta: ␣ 318,  1232 Beta: ␣ 88,  25 Beta: ␣ 100,  1 Beta: ␣ 25  1 Beta: ␣ 183  153 Beta: ␣ 180  149 Beta: ␣ 145  82 Beta: ␣ 72  16
27 27 27 25 25 27 26 26 26
Abbreviations: HT, hemithyroidectomy; TT, total thyroidectomy; RAI, radioactive iodine. a
Parameters for distributions were calculated by assuming a SD of 10% of the estimate if no SD was given in the source.
b
Per recurrent laryngeal nerve.
c
AUS only.
formation was accessed, as per the local policy. The model was created and analyzed using TreeAge Pro 2012 (TreeAge Software Inc) and STATA 12 (StataCorp).
Setting The analysis was performed from the US third-party payer (Medicare) and Canadian healthcare system perspectives (ie, societal costs such as productivity losses were not included). Probabilities for outcomes were obtained from specific literature searches (Table 1). If more than one relevant study was identified, then a weighted pooled proportion was calculated using a random-effects model. Future costs and effectiveness were discounted at a 3% rate, but this was varied from 0% to 5% on sensitivity analyses.
Outcomes Costs Costs were expressed as 2013 US$ or CAN$ (1 US$ ⫽ 0.77 CAN$ through purchasing power parity) (18) (Table 2). The costs of the GEC and the GMP were obtained from the respective manufacturers, expressed as US$ and converted into CAN$, because no Canadian-specific pricing is available. The remaining costs of the surgery, adjuvant treatment, and treatment of complications were each derived individually from the US or Canadian perspective. All costs include physician fees. US costs were
obtained from the Healthcare Cost and Utilization Project from the Agency of Healthcare Research and Quality (19) using average 2011 Medicare costs for the appropriate ICD-9 procedure codes, and from the 2013 Medicare physician and clinical laboratory fee schedules (2013 conversion factor ⫽ US$25.00) (20). Canadian (province of Quebec) costs were obtained from the Canadian Institute for Health Information case-mix groups (21), which are similar to diagnosis-related groups. These costs represent the mean overall cost of the procedure, including the operation and subsequent hospitalization. Other costs and physician fees were obtained from Quebec provincial cost manuals (22, 23). Drug costs were based on US (24) and Quebec provincial (23) wholesale costs.
Effectiveness QALYs were used as the main effectiveness measure. QALYs are calculated by multiplying the utility of a health state by the time spent in that health state. Utilities (also known as quality of life weights) are valued from a scale of 0 (death) to 1 (perfect health). Utilities for each of the health states in our model were taken from three different studies (Table 1) (25–27). Costs and effectiveness measures were incorporated into a single summary measure as the incremental cost-effectiveness ratios.
Sensitivity analysis Deterministic and probabilistic sensitivity analyses were performed to account for parameter uncertainty. Deterministic sen-
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 02 September 2014. at 01:23 For personal use only. No other uses without permission. . All rights reserved.
2678
Lee et al
Molecular Testing for Thyroid Nodules
J Clin Endocrinol Metab, August 2014, 99(8):2674 –2682
Table 2. Costs of Parameters Included in the Model From the US Third-Party Payer and Canadian Healthcare System Perspectives United States, 2013 $US
GEC GMP Thyroid lobectomy Total thyroidectomy Completion thyroidectomy Neck dissection Radioactive iodine ablation (whole body I131 scan, I131 ablation, 1 inpatient day) Yearly cost of surveillance (TSH, thyroglobulin, anti-thyroglobulin antibodies, neck ultrasound, and physical examination) Yearly cost of follow-up of benign nodules (neck ultrasound, physical examination) Yearly cost of TSH suppression (assuming daily dose of levothyroxine 100 g) Yearly cost of treatment of hypoparathyroidism (assuming daily dose of calcitriol 25 g and calcium/vitamin D 1000 mg/800 U)
Canada, 2013 $CAN
Estimatea (rangeb)c
Source or Ref.
Estimatea (rangeb)c
Source or Ref.
3500 (1750 –7000) 850 (425–1700) 9042 (2521–18 084) 10 155 (5078 –20 310) 10 364 (5182–20 728) 16 729 (8365–33 458) 451 (226 –902)
Manufacturer Manufacturer 19, 20 19, 20 19, 20 19, 20 20
4340 (2170 – 8680) 1054 (527–2108) 4205 (2103– 8410) 4483 (2242– 8966) 4359 (2180 – 8718) 6120 (3060 –12 240) 1580 (790 –3160)
Manufacturer Manufacturer 21, 22 21, 22 21, 22 21, 22 22, 23
309 (155– 618)
20
220 (110 – 440)
22, 23
242 (121– 484)
20
136 (68 –272)
22, 23
147 (74 –294)
24
22 (11– 44)
23
242 (121– 484)
24
453 (227–906)
23
a
All costs include physicians’ fees.
b
Ranges were calculated using 50% of the base estimate for the lower limit and 200% for the upper limit.
␥-Distributions were fitted around each cost estimate using ␣ ⫽ 1 and  ⫽ base case cost as parameters, except for the costs of the GEC and the GMP, which were not subject to probabilistic uncertainty. c
sitivity analysis was performed by varying one or more variables at a time across a specified range of values. Ranges were derived from the literature (directly or from pooled analyses). Microsimulation models inherently perform probabilistic sensitivity analysis because parameters for each simulated patient are randomly drawn from the distributions around each variable. The appropriate distributions were fitted around each variable (normal distribution for relative risks,  distribution for probabilities and utilities, and ␥ distribution for costs), which were derived from the literature. The 95% confidence intervals around the point estimates were obtained from the 2.5th and 97.5th percentiles. Uncertainty around cost-effectiveness results was represented through cost-effectiveness acceptability curves, which represent the proportion of times (out of the total number of simulated patients) that each option is optimal for a given cost/ QALY threshold.
Results The mean age of the simulated patients was 54.4 years (SD, 21.3), and 19% were male. Results of the cost-effectiveness analyses from the US and Canadian perspectives are reported in Table 3. From the US perspective, routine use of GEC with selective use of GMP resulted in the lowest cost and the most QALYs (ie, dominant) compared to the other four strategies. From the Canadian perspective, this strategy was dominant over the routine use of GEC
only, but it resulted in an additional CAN$24 030 per QALY gained when compared to standard management. From both perspectives, the routine use of the GEC-only strategy was the most costly. As well, the routine use of GMP and the routine use of GMP with selective use of GEC strategies were dominated (ie, more costly and less effective) from both perspectives. There was no difference in life expectancy across strategies from either perspective (Table 3). Similarly, there was minimal difference in QALYs between strategies. The cost-effectiveness results were most sensitive to variations in the cost of the GEC and GMP (Figure 2, A and B). Sensitivity analyses demonstrated that in order for the routine GEC ⫹ selective GMP strategy to be less costly than standard management, the GEC needed to cost less than US$5036 and the GMP less than US$1609 from the US perspective. From the Canadian perspective, this strategy was the least costly only if the GEC cost less than CAN$2215 and the GMP less than CAN$721. From both perspectives, the routine GMP and the routine GMP ⫹ selective GEC strategies were never the least costly option across the range of values. From the US perspective, the routine GEC ⫹ selective GMP strategy would be the least costly only if the cost of a thyroid lobectomy exceeded US$6686. Other variables tested had minimal impact (data not shown).
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 02 September 2014. at 01:23 For personal use only. No other uses without permission. . All rights reserved.
doi: 10.1210/jc.2014-1219
jcem.endojournals.org
2679
Table 3. Cost-Effectiveness Results From Both the US Third-Party Payer and Canadian Healthcare System Perspectives
US third-party payer perspective, 2013 US$ Standard management Routine GEC Routine GEC with selective GMP Routine GMP Routine GMP with selective GEC Canadian healthcare system perspective, 2013 CAN$ Standard management Routine GEC Routine GEC with selective GMP Routine GMP Routine GMP with selective GEC
Mean Life Expectancy, y (95% CI)
Mean Cost (95% CIa)
27.64b (27.24, 28.04)
22 385 (22 178, 22 593)
27.41 (27.02, 27.82) 27.53 (27.13, 27.93)
22 586 (22 331, 22 846) 20 506 (20 287, 20 732)
⫹201 (56, 348) ⫺2080 (⫺2285, ⫺1875)
27.65 (27.24, 28.05) 27.45 (27.05, 27.84)
21 588 (21 396, 21 776) 21 267 (21 040, 21 491)
28.52e (28.11, 28.93) 28.57 (28.15, 28.98) 28.78 (28.37, 29.20) 28.28 (27.87, 28.69) 28.44 (28.03, 28.86)
Incremental Cost (95% CI)
Mean QALYs (95% CI)
Incremental QALYs (95% CI)
ICER, $/QALY
17.23 (17.06, 17.39) 17.28 (17.12, 17.45)
⫹0.12 (0.01, 0.23) ⫹0.03 (⫺0.05, 0.12)
16 750 Dominantc,d
⫹1082 (780, 1383) ⫺320 (⫺562, ⫺78)
17.12 (16.97, 17.29) 17.09 (16.93, 17.25)
⫺0.26 (⫺0.36, ⫺0.15) ⫹0.10 (⫺0.01, 0.22)
Dominated Dominated
11 050 (10 942, 11 157) 13 814 (13 685, 13 941) 13 453 (13 335, 13 573)
⫹2764 (2692, 2835) ⫺360 (⫺464, ⫺257)
17.18 (17.02, 17.35) 17.25 (17.09, 17.41) 17.28 (17.12, 17.45)
⫹0.07 (⫺0.04, 0.19) ⫹0.05 (⫺0.03, 0.14)
39485 24 030,c Dominantd
11 851 (11 738, 11 963) 13 640 (13 519, 13 756)
⫺1602 (⫺1786, ⫺1418) ⫹1788 (1644, 1933)
17.02 (16.86, 17.19) 17.12 (16.96, 17.29)
⫺0.15 (⫺0.26, ⫺0.04) ⫺0.04 (⫺0.15, 0.07)
Dominated Dominated
17.11 (16.94, 17.27)
Abbreviations: CI, confidence interval; ICER, incremental cost-effectiveness ratio; Dominant, less costly and more effective; Dominated, more costly and/or less effective. a
CIs around mean costs derived from bootstrapped estimates. Other CIs are normal distribution-based.
b
P value ⫽ .552 for the differences in life expectancy between strategies from the US perspective.
c
Compared to standard management.
d
Compared to routine GEC strategy.
e
P value ⫽ .895 for the differences in life expectancy between strategies from the Canadian perspective.
The results were also sensitive to variations in the probability of malignancy from both perspectives. The interaction between the costs of the GEC and the probability of malignancy demonstrated that the routine GEC strategy was the least costly strategy only in cases when the cost of the GEC and probability of malignancy were both low, although the
threshold values were different for each perspective (Figure 2, C and D). Again, the routine GMP and the routine GMP ⫹ selective GEC strategies were never the least costly option across the range of values. Cost-effectiveness acceptability curves are shown in Figure 3, A and B. From the US perspective, the routine GEC ⫹ selective GMP strategy had the highest probability of cost-effectiveness across all cost/QALY thresholds, whereas from the Canadian perspective, standard management was most likely to be cost effective. However, there was significant uncertainty around these results from both perspectives because none of the strategies had a probability of cost-effectiveness higher than 35% at any cost/ QALY threshold.
Discussion Figure 2. Relationship between variations in cost of GEC and GMP on incremental strategy costs from the US third-party payer (A) and Canadian healthcare system (B) perspectives; and between variations in the cost of the GEC and risk of malignancy in nodule from the US thirdparty payer (C) and Canadian healthcare system (D) perspectives.
Thyroid nodules with AUS represent an important clinical dilemma. Novel molecular diagnostic markers, such as gene expression and gene muta-
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 02 September 2014. at 01:23 For personal use only. No other uses without permission. . All rights reserved.
2680
Lee et al
Molecular Testing for Thyroid Nodules
J Clin Endocrinol Metab, August 2014, 99(8):2674 –2682
Figure 3. Cost-effectiveness acceptability curves from the US third-party payer (A) and Canadian healthcare system (B) perspectives.
tion testing, may help identify nodules that are more likely to be benign or malignant and help avoid unnecessary surgery and the risks associated with thyroidectomy (28). However, these novel markers are costly, and their costeffectiveness has yet to be fully evaluated. In the present study, a strategy combining routine GEC and selective GMP use for AUS thyroid nodules was most likely to be cost effective in the US setting, whereas standard management was the strategy with the highest probability of costeffectiveness from a Canadian perspective. Although the cost-effectiveness results from both perspectives appear favorable for the strategy combining routine GEC ⫹ selective GMP, there was considerable uncertainty around the estimates, as demonstrated in the cost-effectiveness acceptability curves (Figure 3, A and B). This was also reflected in deterministic sensitivity analyses, which reported important changes in the interpretation of the cost-effectiveness results based on variations in the cost of the GEC, diagnostic thyroid lobectomy, and the probability of malignancy. There were also important differences between the US and Canadian perspectives. In the United States, there were cost savings because of the high cost of thyroid lobectomy relative to the cost of the molecular markers, whereas in Canada, the low cost of surgery relative to the molecular markers resulted in higher costs associated with their use. Favorable cost-effectiveness results are obtained with current US pricing, but the price of both tests will have to be lower in Canada before molecular marker diagnostics becomes cost saving. Also, the addition of gene mutation testing to routine gene expression testing becomes less costly overall as the probability of malignancy becomes higher because more patients with a positive GEC result would receive singlestage total thyroidectomy, as long as the cost of the GEC remains (relatively) low. Differences in the cost of the competing strategies were more important compared to differences in their effective-
ness because there were minimal to nonexistent differences in life expectancy or QALYs between the strategies (slight differences between the US and Canadian settings were due to differences in age-specific mortality estimates between the two countries). This can be explained by the fact that the risks of surgical complications that would have an important negative impact on quality of life, such as RLN, and thyroid cancer-related mortality were low (29, 30). Thus, the cost-effectiveness results were mostly dependent on the minimization of unnecessary surgery. The use of gene mutation testing in nodules with AUS cytology is controversial. Previous studies have reported that testing for gene mutations was highly specific (with high positive predictive value) for malignancy but had low sensitivity. Therefore, gene mutation testing is ideal to determine the extent of surgery in patients with a high suspicion of malignancy because those patients who test positive for a gene mutation can avoid a two-stage operation (diagnostic thyroid lobectomy, followed by completion thyroidectomy if malignant) and proceed directly to total thyroidectomy. However, the diagnostic characteristics of gene mutation testing (ie, low sensitivity and negative predictive value) do not allow for the accurate identification of nodules with relatively low probability of malignancy (6). Furthermore, there are data reporting that a high proportion of follicular variant of papillary thyroid cancers do not express the BRAF mutation and therefore may not register positive on GMP testing (31, 32). Therefore, patients with a negative GMP result should still undergo diagnostic lobectomy, given the increased risk of false negatives. The selective use of a GEC after a negative GMP result may reduce the number of false negatives and unnecessary surgery, but these two strategies (routine GMP only and routine GMP ⫹ selective GEC) were not costeffective from either perspective. There have been previous studies that investigated the cost-effectiveness of gene expression and gene mutation
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 02 September 2014. at 01:23 For personal use only. No other uses without permission. . All rights reserved.
doi: 10.1210/jc.2014-1219
testing for thyroid nodules with indeterminate cytology. Li et al (10) performed a model-based economic evaluation of routine GEC use compared to standard practice in the United States over a 5-year time horizon and reported cost savings with a small increase in QALYs. Yip et al (9) also used a model-based approach to perform a cost analysis comparing routine GMP use to standard practice in the United States and reported favorable results. No studies have performed an economic evaluation alongside a clinical trial or using individual-patient data. There are some important differences between these studies and our results. In the present study, routine GEC was associated with increased costs, whereas both previous studies reported cost savings (10). This discrepancy may be explained in part by differences in model parameters, model type and structure, and time horizon. For example, Li et al (10) modeled cost-effectiveness results over a 5-year time horizon, which may not be long enough for a slowly progressive disease such as thyroid cancer. Furthermore, these two studies used less sophisticated models (with more constraints) than the present study. Although all model-based evaluations are simplified representations of reality, a less constrained model that more accurately mimics the natural history of a disease, such as microsimulation, is more likely to provide realistic results (33). Lastly, the present study evaluated two additional combination strategies of routine GEC ⫹ selective GMP and routine GMP ⫹ selective GEC, and performed the economic evaluation from multiple perspectives. There are also some preliminary data to suggest that papillary microcarcinomas (less than 1 cm) diagnosed on FNA can be observed, and resected only if they progress, with similar outcomes as outright resection (34, 35). Although this approach may certainly affect the cost-effectiveness of the management strategies that were evaluated in the present study, it is not directly germane to our objectives and data obtained because our target population has not yet been diagnosed with thyroid cancer and is still undergoing diagnostic investigations. This study has limitations. As with any model-based economic evaluation, the inputs for model parameters are based on data from other studies and are subject to the limitations of the source studies themselves. The diagnostic performance characteristics of the GEC and the GMP were based on two studies (7, 8). It is possible that the diagnostic performance characteristics of either test may change as more validation studies are performed. Other model inputs were obtained largely from single-center observational studies. Also, the results of this study may not be generalizable across all practice settings because our results demonstrate that the cost-effectiveness of the competing strategies was highly sensitive to variations in the
jcem.endojournals.org
2681
cost of important parameters. This is especially important given that previous studies have shown wide variations in the cost of care depending on the Medicare region (36). There were also certain assumptions that were made regarding treatment options that remain controversial, such as routine radioactive iodine remnant ablation and the choice of total thyroidectomy for definitive treatment (37). In these cases, we modeled management recommendations based on ATA guidelines (3). In summary, the results of this economic evaluation suggest that a strategy combining routine gene expression testing with selective gene mutation testing may be cost effective in the management of patients with AUS thyroid nodules. However, this decision was subject to considerable uncertainty because the cost-effectiveness of this strategy was different in the United States than in Canada and was highly sensitive to important model parameters, such as the cost of the GEC, thyroid lobectomy, and the probability of malignancy in the nodule. In order for molecular diagnostic markers to be more cost effective in a setting outside the United States, the cost of these tests needs to be revised.
Acknowledgments Address all correspondence and requests for reprints to: Lawrence Lee, MD, MSc, Steinberg-Bernstein Centre for Minimally Invasive Surgery and Innovation, Department of Surgery, McGill University Health Centre, 1650 Cedar Avenue, E19-125, Montreal, QC H3G 1A4, Canada. E-mail: [email protected]. Disclosure Summary: The authors have nothing to declare.
References 1. Davies L, Welch HG. Increasing incidence of thyroid cancer in the United States, 1973–2002. JAMA. 2006;295:2164 –2167. 2. Grodski S, Brown T, Sidhu S, et al. Increasing incidence of thyroid cancer is due to increased pathologic detection. Surgery. 2008;144: 1038 –1043. 3. Cooper DS, Doherty GM, Haugen BR, et al. Revised American Thyroid Association management guidelines for patients with thyroid nodules and differentiated thyroid cancer. Thyroid. 2009;19:1167– 1214. 4. Cibas ES, Ali SZ. The Bethesda system for reporting thyroid cytopathology. Am J Clin Pathol. 2009;132:658 – 665. 5. Wang CC, Friedman L, Kennedy GC, et al. A large multicenter correlation study of thyroid nodule cytopathology and histopathology. Thyroid. 2011;21:243–251. 6. Xing M, Haugen BR, Schlumberger M. Progress in molecular-based management of differentiated thyroid cancer. Lancet. 2013;381: 1058 –1069. 7. Alexander EK, Kennedy GC, Baloch ZW, et al. Preoperative diagnosis of benign thyroid nodules with indeterminate cytology. N Engl J Med. 2012;367:705–715. 8. Nikiforov YE, Ohori NP, Hodak SP, et al. Impact of mutational
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 02 September 2014. at 01:23 For personal use only. No other uses without permission. . All rights reserved.
2682
9.
10.
11. 12.
13.
14.
15.
16.
17.
18.
19.
20.
21. 22.
23.
24. 25.
Lee et al
Molecular Testing for Thyroid Nodules
testing on the diagnosis and management of patients with cytologically indeterminate thyroid nodules: a prospective analysis of 1056 FNA samples. J Clin Endocrinol Metab. 2011;96:3390 –3397. Yip L, Farris C, Kabaker AS, et al. Cost impact of molecular testing for indeterminate thyroid nodule fine-needle aspiration biopsies. J Clin Endocrinol Metab. 2012;97:1905–1912. Li H, Robinson KA, Anton B, Saldanha IJ, Ladenson PW. Costeffectiveness of a novel molecular test for cytologically indeterminate thyroid nodules. J Clin Endocrinol Metab. 2011;96:E1719 – E1726. Bodenheimer T. High and rising health care costs. Part 1: seeking an explanation. Ann Intern Med. 2005;142:847– 854. Yeh MW, Demircan O, Ituarte P, Clark OH. False-negative fineneedle aspiration cytology results delay treatment and adversely affect outcome in patients with thyroid carcinoma. Thyroid. 2004; 14:207–215. Najafzadeh M, Marra CA, Lynd LD, Wiseman SM. Cost-effectiveness of using a molecular diagnostic test to improve preoperative diagnosis of thyroid cancer. Value Health. 2012;15:1005–1013. Bilimoria KY, Bentrem DJ, Ko CY, et al. Extent of surgery affects survival for papillary thyroid cancer. Ann Surg. 2007;246:375–381; discussion 381–384. Arias E. United States life tables, 2008. National Vital Statistics Reports. Vol 61, no. 3. http://www.cdc.gov/nchs/data/nvsr/nvsr61/ nvsr61_03.pdf. Accessed July 12, 2013. Life Tables, Canada, Provinces and Territories 2007 to 2009. http://www.statcan.gc.ca/pub/84-537-x/84-537-x2013003-eng. htm. Accessed July 1, 2013. Rouxel A, Hejblum G, Bernier MO, et al. Prognostic factors associated with the survival of patients developing loco-regional recurrences of differentiated thyroid carcinomas. J Clin Endocrinol Metab. 2004;89:5362–5368. Organization for Economic Cooperation and Development. 2008 PPP Benchmark results. http://stats.oecd.org/Index.aspx?DataSetCode⫽CPL. Accessed January 6, 2013. Agency for Healthcare Research and Quality. Healthcare Cost and Utilization Project (HCUP.net). http://hcupnet.ahrq.gov/. Accessed October 15, 2013. Centers for Medicare & Medicaid Services. Physician fee schedule. http://www.cms.gov/Medicare/Medicare-Fee-for-Service-Payment/ PhysicianFeeSched/. Accessed October 22, 2013. Canadian Institute for Health Information. Case mix groups (CMG⫹). http://www.cihi.ca/casemix. Accessed March 30, 2013. Régie de l’assurance maladie du Québec. Manuel de facturation. http:// www.ramq.gouv.qc.ca/fr/professionnels/medecins-specialistes/manuels/ Pages/facturation.aspx. Accessed May 1, 2013. Ministère de la Santé et des Services sociaux. Normes et pratiques de gestion pour les centres hospitaliers. http://msssa4.msss.gouv.qc.ca/ fr/document/d26ngest.nsf/lcaCH?OpenView. Accessed October 22, 2013. Medi-Span. http://www.medispan.com/. Accessed October 30, 2013. Esnaola NF, Cantor SB, Sherman SI, Lee JE, Evans DB. Optimal
J Clin Endocrinol Metab, August 2014, 99(8):2674 –2682
26.
27.
28.
29.
30.
31.
32.
33.
34.
35.
36.
37.
38.
39.
40.
treatment strategy in patients with papillary thyroid cancer: a decision analysis. Surgery. 2001;130:921–930. Mernagh P, Suebwongpat A, Silverberg J, Weston A. Cost-effectiveness of using recombinant human thyroid-stimulating hormone before radioiodine ablation for thyroid cancer: the Canadian perspective. Value Health. 2010;13:180 –187. Kebebew E, Duh QY, Clark OH. Total thyroidectomy or thyroid lobectomy in patients with low-risk differentiated thyroid cancer: surgical decision analysis of a controversy using a mathematical model. World J Surg. 2000;24:1295–1302. Alexander EK, Schorr M, Klopper J, et al. Multicenter clinical experience with the Afirma gene expression classifier. J Clin Endocrinol Metab. 2014;99:119 –125. Rosato L, Avenia N, Bernante P, et al. Complications of thyroid surgery: analysis of a multicentric study on 14,934 patients operated on in Italy over 5 years. World J Surg. 2004;28:271–276. Hundahl SA, Fleming ID, Fremgen AM, Menck HR. A National Cancer Data Base report on 53,856 cases of thyroid carcinoma treated in the U.S., 1985–1995. Cancer. 1998;83:2638 –2648. Lim JY, Hong SW, Lee YS, et al. Clinicopathologic implications of the BRAF(V600E) mutation in papillary thyroid cancer: a subgroup analysis of 3130 cases in a single center. Thyroid. 2013;23:1423– 1430. Trovisco V, Soares P, Preto A, et al. Type and prevalence of BRAF mutations are closely associated with papillary thyroid carcinoma histotype and patients’ age but not with tumour aggressiveness. Virchows Arch. 2005;446:589 –595. Weinstein MC. Recent developments in decision-analytic modelling for economic evaluation. Pharmacoeconomics. 2006;24:1043– 1053. Sugitani I, Toda K, Yamada K, Yamamoto N, Ikenaga M, Fujimoto Y. Three distinctly different kinds of papillary thyroid microcarcinoma should be recognized: our treatment strategies and outcomes. World J Surg. 2010;34:1222–1231. Ito Y, Miyauchi A, Kihara M, Higashiyama T, Kobayashi K, Miya A. Patient age is significantly related to the progression of papillary microcarcinoma of the thyroid under observation. Thyroid. 2014; 24:27–34. Fisher ES, Wennberg DE, Stukel TA, Gottlieb DJ, Lucas FL, Pinder EL. The implications of regional variations in Medicare spending. Part 1: the content, quality, and accessibility of care. Ann Intern Med. 2003;138:273–287. McLeod DS, Sawka AM, Cooper DS. Controversies in primary treatment of low-risk papillary thyroid cancer. Lancet. 2013;381: 1046 –1057. Mazzaferri EL, Jhiang SM. Long-term impact of initial surgical and medical therapy on papillary and follicular thyroid cancer. Am J Med. 1994;97:418 – 428. Landerholm K, Wasner AM, Järhult J. Incidence and risk factors for injuries to the recurrent laryngeal nerve during neck surgery in the moderate-volume setting [published online January 9, 2014]. Langenbecks Arch Surg. doi:10.1007/s00423-013-1154-6. Giordano D, Valcavi R, Thompson GB, et al. Complications of central neck dissection in patients with papillary thyroid carcinoma: results of a study on 1087 patients and review of the literature. Thyroid. 2012;22:911–917.
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 02 September 2014. at 01:23 For personal use only. No other uses without permission. . All rights reserved.
