Int J Colorectal Dis (2014) 29:1061–1068 DOI 10.1007/s00384-014-1917-8
ORIGINAL ARTICLE
The impact of pathologic nodal status on survival following neoadjuvant chemoradiation for locally advanced rectal cancer Jonathan M. Hernandez & Whalen Clark & Jill Weber & William J. Fulp & Lauren Lange & David Shibata
Accepted: 2 June 2014 / Published online: 27 June 2014 # Springer-Verlag Berlin Heidelberg 2014
Abstract Purpose For patients with locally advanced rectal cancer, the accuracy rates of preneoadjuvant therapy nodal staging and potential nodal downstaging make the prognostic significance of nodal status unclear. We therefore sought to review our experience in order to better understand the impact of clinical and pathologic nodal status upon patient outcomes. Methods 174 patients were identified as having undergone neoadjuvant chemoradiation and resection for rectal cancer. For analytic purposes, patients were grouped into four nodal categories (uN 0·pN 0, uN 0·pN +, uN + ·pN 0, and uN + ·pN +). Univariate and multivariate analyses were performed. Results 104 men and 70 women of median age 60 years (29– 85 years) were followed for a median of 31 months (1– 121 months). Nodal staging was available for 129 patients, with a median of 8 lymph nodes (range 0–39) evaluated. Disease recurred in 3 of 41 (7 %) uN 0·pN 0, 10 of 52 (20 %) uN +·pN 0, 7 of 18 (41 %) uN 0·pN +, and 6 of 17 (35 %) uN + ·pN + patients. Th os e pat i en t s ha vi ng no dal downstaging (uN +·pN 0) experienced superior overall survival (p=0.03). Only pathologic nodal status was a significant predictor of both disease-free and overall survival in multivariate modeling. Adjuvant chemotherapy did not impact disease-free or overall survival for patients with pN0. Conclusions Pathologic nodal status may represent a superior predictor of survival for patients with local advanced rectal
J. M. Hernandez : W. Clark : J. Weber : L. Lange : D. Shibata (*) Department of Gastrointestinal Oncology, H. Lee Moffitt Cancer Center and Research Institute, Tampa, FL, USA e-mail: [email protected] W. J. Fulp Biostatistics Core, H. Lee Moffitt Cancer Center and Research Institute, Tampa, FL, USA
cancers. Our findings may have potential implications for the application of adjuvant therapy. Keywords Rectal cancer . Lymph nodes . Neoadjuvant therapy . Resection . EUS
Introduction Rectal cancers extending through the rectal wall or involving locoregional lymph nodes have historically been associated with significant rates of local relapse [1]. The initial application of postoperative [2] and subsequently, preoperative chemoradiation in combination with radical resection has been shown to result in improved outcomes [3]. Consequently, the current preferred standard of care for patients diagnosed with rectal adenocarcinoma and tumor invasion beyond the muscularis propria (T3 or T4) or preoperative nodal involvement (N+) by imaging criteria is neoadjuvant chemoradiation followed by radical resection and subsequent adjuvant chemotherapy [4]. Sole reliance on preoperative imaging modalities for stage assignment has given rise to concerns that a significant number of patients will be overtreated or undertreated. Given that endorectal ultrasound (EUS), CT and pelvic MRI are associated with significant false-negative rates, and that inflammation, scar, stool, and other artifacts can cause false positives during pretreatment scanning [5]; the significance of postoperative pathologic nodal status following rectal resection is of interest. We sought to evaluate whether pathologic nodal stage, which is routinely reported for all patients undergoing resection, can be used as a prognostic variable following multimodality treatment of rectal cancer. We hypothesized that pathologic nodal status would be an important predictor of outcome and may be potentially useful for determining the application of adjuvant therapy in such cases.
1062
Methods Comprehensive clinicopathologic data were collected from 1998 to 2008 for this IRB-approved study. Patients were identified based upon their diagnosis of stage II or III rectal adenocarcinoma and completion of neoadjuvant chemoradiation followed by radical surgical resection. All patients were locally staged prior to receipt of neoadjuvant therapy with endorectal ultrasound and, in some cases, pelvic MRI. Abnormal nodes meeting imaging criteria for positivity were generally not confirmed with biopsy. Neoadjuvant therapy consisted primarily of 5-FU-based chemotherapy, with concurrent external beam radiation (50.4 Gy). Adjuvant chemotherapy consisted of 5-FU, leucovorin, and oxaliplatin (FOLFOX). Disease-free and overall survival analyses were performed using the Kaplan-Meier method and Cox proportional hazard models. Univariate analyses were performed using Chi-squared test, Fisher’s exact test, or ANOVA where appropriate. Significant variables by univariate analyses were considered for multivariate modeling. The final multivariate models were developed by backward elimination using Akaike information criterion. For comparison, patients were grouped into four paired preoperative and postoperative nodal categories (uN 0·pN 0, uN 0·pN +, uN + ·pN 0, and uN + ·pN +).
Int J Colorectal Dis (2014) 29:1061–1068 Table 1 Characteristics of 174 rectal cancer patients
Percent
106 68 60.5
60.9 39.0 12.9
White 165 Nonwhite 7 Unknown 2 Tumor stage (at EUS) T1 1 T2 12 T3 124 T4 6 Nodal stage (at EUS) N0 63 N+ 73 Nx 7 Tumor stage (at resection) T0 42 Tis 1 T1 10 T2 51 T3 64 T4 6 Nodal stage (at resection) N0 117 N1 32 N2 12 Nx 13 Type of surgery LAR 134 APR 34 Other 6
94.8 4.0 1.1
Gender Male Female Age Race
Results We identified 174 patients with locally advanced rectal cancer treated at our institution (104 men and 70 women) with a median age of 60 years (range 29–85). Following completion of neoadjuvant chemoradiation, 140 patients underwent low anterior resection and 34 underwent abdominoperineal resection (APR). Following resection, 90 % of patients completed adjuvant chemotherapy. Comprehensive demographic data is shown in Table 1. Median disease-free and overall survival for the cohort was 66.7 and 73.8 months, respectively. Univariate analyses were undertaken to determine the impact of various conventional parameters upon disease-free and overall survival (Table 2). Interestingly, in our patient population, “T” downstaging had no demonstrable impact upon disease-free or overall survival (Fig. 1a, b). Pretreatment (radiologic) and postoperative (pathologic) nodal staging was available for 129 patients. The median number of lymph nodes evaluated in patients with pretreatment and postoperative nodal staging was 8 (range 0–39).
Number
0.7 8.4 86.7 4.2 44.1 51 4.9
24.1 0.6 5.8 29.3 36.8 3.4 67.2 18.4 6.9 7.5 77 19.5 3.5
metastases on pathologic evaluation. Median disease-free and overall survival for patients with nodal staging of uN 0·pN 0 were not met (Fig. 2a, b). Of the 42 patients staged uN 0·pN 0, 3 (7 %) patients experienced disease recurrence (Table 3). On univariate analysis, patients without lymph node metastases on final pathology had significantly prolonged disease-free survival as compared to patients with lymph node metastases (Fig. 2a, p=0.02).
Nodal status uN 0·pN 0 Forty-two patients did not have evidence of lymph node metastases on preoperative imaging and ultimately no nodal
uN + ·pN 0 Fifty-two patients had evidence of lymph node metastases on preoperative imaging but ultimately did not have nodal metastases on pathologic evaluation. Median disease-free and
Int J Colorectal Dis (2014) 29:1061–1068
1063
Table 2 Univariate model for disease-free and overall survival Variable
Disease-free survival (p value)
Overall survival (p value)
Grade Perineural invasion Margin (circumferential resection margin) Circumferential tumor APR T stage Nodal category EUS nodal stage Pathologic nodal stage Adjuvant chemotherapy
0.23 0.16 0.03
0.22 0.35 0.41
0.77 0.11 0.06 0.003 0.92 0.004 0.90
0.70 0.14 0.26 0.05 0.70 0.05 0.20
uN 0·pN +
1.00
Eighteen patients did not have evidence of lymph node metastases on preoperative imaging but ultimately had nodal
0.50
0.75
T-Stage 0 n=42
0.25
T-Stage 1-4 n=131
p=0.0664
0.00
Probability of Survival
a
0
1
2
3
4
5
T-Stage 0 n=42
0.50
0.75
b
1.00
Time (Yrs)
0.25
T-Stage 1-4 n=131
p=0.3167
0.00
Probability of Survival
Fig. 1 Disease-free (a) and overall survival (b) stratified by tumor stage at resection
overall survival for patients with nodal staging of uN +·pN0 were 73.8 and 85.4 months (Fig. 2a, b), respectively. Of the 52 patients staged as uN +·pN 0, 10 (20 %) patients experienced disease recurrence (Table 3). On univariate analysis, patients without lymph node metastases on final pathology had a significantly prolonged disease-free survival as compared to patients with lymph node metastases (Fig. 2a, p=0.02). Patients experiencing nodal downstaging (uN +·pN 0) experienced superior overall survival as compared to patients experiencing nodal upstaging (uN 0·pN +) (Fig. 2c, p=0.03).
0
1
2 Time (Yrs)
3
4
5
1064
uN0,pN0 (n=42)
0.50
uN+,pN0 (n=52) uN+,pN+ (n=17)
0.25
uN0,pN+ (n=18)
p=0.0222
0.00
Probability of Survival
0.75
1.00
a
b
1
2 Time (Yrs)
3
4
5
1.00
0
0.75 0.25
0.50
uN0,pN0 (n=42) uN+,pN+ (n=17)
uN0,pN+ (n=18)
0.00
Probability of Survival
uN+,pN0 (n=52)
p=0.2167 0
1
2
3
4
5
c
1.00
Time (Yrs)
0.75
uN+,pN0 n=52
uN0,pN+
0.25
0.50
n=18
p=0.03
0.00
Probability of Survival
Fig. 2 Disease-free (a) and overall survival (b, c) stratified by pretherapy (u) and posttherapy (p) nodal groups
Int J Colorectal Dis (2014) 29:1061–1068
0
1
2 Time (Yrs)
3
4
5
Int J Colorectal Dis (2014) 29:1061–1068
1065
Table 3 Locations of first recurrence stratified by nodal group Locations of first recurrence Nodal group uN 0⋅pN 0a uN+⋅pN 0 uN 0⋅pN + a uN+⋅pN+ a
Number 3 10 7 6
Local 1 2 1 2
Liver 1 3 3 2
Lung 1 3 3 1
Other 1 2 3 1
Some patients had synchronous recurrences in more than one location
metastases on pathologic evaluation. Median disease-free and overall survival for patients with nodal staging of uN 0·pN + were 38.1 and 55.7 months (Fig. 2a, b). Of the 18 patients staged uN 0·pN +, 7 (41 %) patients experienced disease recurrence (Table 3). On univariate analysis, patients with lymph node metastases on final pathology had significantly abbreviated disease-free survival as compared to patients without lymph node metastases (Fig. 2a, p=0.02). Patients experiencing nodal upstaging (uN 0·pN +) experienced abbreviated overall survival as compared to patients experiencing nodal downstaging (uN +·pN 0) (Fig. 2c, p=0.03). uN +·pN + Seventeen patients had evidence of lymph node metastases on preoperative imaging and ultimately had persistent nodal metastases on pathologic evaluation. Median disease-free and overall survival for patients with nodal staging of uN +·pN + were 87.1 and 60.5 months (Fig. 2a, b), respectively. Of the 17 patients staged uN +·pN +, 6 (35 %) patients experienced disease recurrence (Table 3). On univariate analysis, patients with lymph node metastases on final pathology had significantly reduced disease-free survival as compared to patients without lymph node metastases (Fig. 2a, p=0.02). Using AIC backward selection, we identified a significant multivariate model to predict disease-free (p=0.0016, Table 4) Table 4 Multivariate model for disease-free survival Variable
Odds ratio
95 % CI
p Value
Age Nodal stage pN0 pN+ Tumor stage T0 T any Type of surgery LAR APR
1.03
0.99–1.05
0.077
– 2.55
– 1.25–5.21
– 0.010
– 2.34
– 0.79–6.90
– 0.12
– 1.89
– 0.87–4.09
– 0.11
and overall survival (p=0.029, Table 5). The variables evaluated in our disease-free and overall survival models included the following: pathologic “T” stage, pathologic “N” stage, age, tumor grade, margin status, circumferential tumor, type of surgery (APR), and adjuvant therapy. Of the individual components of the final disease-free (Table 4) and overall survival (Table 5) models, only pathologic nodal status was a significant prognostic indicator. We also stratified those patients without evidence of lymph node metastases on pathologic examination (uN 0·pN 0 and uN +·pN 0) by receipt of adjuvant chemotherapy. Importantly, adjuvant chemotherapy did not impact disease-free or overall survival (Fig. 3) for patients without evidence of lymph node metastases on final pathology.
Discussion The treatment algorithms for rectal adenocarcinoma are based upon preoperative staging modalities with, at best, a sensitivity of 75 % for nodal metastases with the general assumption that postoperative pathologic evaluation does not necessarily represent the original stage and thus does not accurately provide prognostic information. By default, patients who are generally assigned to neoadjuvant chemoradiation for locally advanced rectal cancer are routinely recommended to also receive adjuvant chemotherapy [4]. Interestingly, for such rectal cancer patients, there are no prospective randomized trials that definitely show a benefit for the application of postoperative adjuvant therapy. Rather, data from colon cancer are extrapolated to this clinical scenario. This approach may potentially augment stage-specific survival given the propensity for false positives, but at what cost? The subset of patients that may truly be node negative and assigned to chemotherapy may unnecessarily suffer the attendant adverse effects of these regimens and may also add to the burden of health care costs. In the present study, we demonstrate that pathologic nodal status is a strong predictor of disease-free and overall survival, irrespective of preoperative stage assignment. Furthermore, we found no benefit from the addition of adjuvant chemotherapy for those patients without evidence of lymphatic Table 5 Multivariate model for overall survival Variable
Odds ratio
95 % CI
p Value
Age Nodal stage pN0 pN+
1.03
0.99–1.06
0.07
– 2.31
– 1.09–4.89
– 0.03
a
Yes
0.25
0.50
No
0.00
Probability of Survival
0.75
Fig. 3 Disease-free (a) and overall survival (b) among patients without evidence of lymph node metastases on pathologic evaluation, stratified by adjuvant chemotherapy
Int J Colorectal Dis (2014) 29:1061–1068
1.00
1066
p=0.9627
0
1
2
3
4
5
b
1.00
Time (Yrs)
0.25
0.50
No
0.00
Probability of Survival
0.75
Yes
p=0.4937
0
1
2
3
4
5
Time (Yrs)
metastases on final pathologic evaluation. Our study suggests that pathologic nodal status may be potentially useful to supplement decision-making with respect to adjuvant treatment. Larger-scale studies of prospective cohorts are certainly warranted. Although the currently available staging modalities are reasonably accurate, they remain imperfect. Based upon the results of large meta-analyses [6], the accuracy of EUS and MRI for determination of T stage is 87 and 71–91 %, respectively. The accuracy of EUS and MRI for determination of “N”
stage is 74 and 45–79 %, respectively. All patients in our study were staged with EUS, and some were additionally imaged with MRI. We recognize the limitations of the current preoperative “T” and “N” staging and the possible tendencies toward overcalling, especially considering that 18 % of nodal metastases occur in lymph nodes less than 5 mm in diameter [7]. Further complicating matters are the lack of accurate predictors of response to neoadjuvant therapy. CEA levels have been suggested in several studies [8–10] as a surrogate marker for response, although cutoff levels have been inconsistent.
Int J Colorectal Dis (2014) 29:1061–1068
Additionally, many studies have attempted to retrospectively correlate immunohistochemical and gene expression biomarkers with response to neoadjuvant therapy [11–18]. At present, the use of biomarkers to predict response remains investigational. Although restaging studies following neoadjuvant therapy are theoretically appealing to help mitigate the inaccuracies of pretreatment stage assignment, none of the currently available examination (DRE, endoscopy) or imaging tools (CT, MR, EUS, PET) are accurate in the setting of fibrosis and inflammation caused by radiation [19]. For example, a prospective study comparing postchemoradiation staging using CT, MR, and EUS to histopathologic findings demonstrated accuracy rates of 37–62, 34–68, and 27–65 %, respectively; most of which was due to overstaging [20]. We do not routinely locally restage patients following neoadjuvant therapy for the reasons stated above, although physical and endoscopic examination prior to resection is generally advised. We did not find any correlation between primary tumor (T) downstaging and survival endpoints. In comparing total primary tumor regression (pT0) to any residual primary tumor (pT+), we did not find any significant correlation with diseasefree or overall survival. Similar results were demonstrated in a multivariate model constructed on a cohort of 153 patients with advanced rectal cancer, in which tumor regression grade is not associated with survival endpoints [21]. Interestingly, other studies have demonstrated a correlation between poor response to neoadjuvant therapy and increased rates of local recurrence [22]. Perhaps, these disparate findings may be the result of ill-defined points of optimal cutoff for separating good and poor response, as has been recently suggested [23]. This problem has been recently addressed by Dhadda et al. [24], who developed a scoring system incorporating cut point tumor regression grades (TRGs), although large-scale validation will be required. Although there is no minimum lymph node retrieval suggested for rectal cancer operations (as opposed to colon cancer), a mean of eight lymph nodes was evaluated for the patients in our study, which is comparable to other series with similar patients [25, 26]. Interestingly, retrieval of
ORIGINAL ARTICLE
The impact of pathologic nodal status on survival following neoadjuvant chemoradiation for locally advanced rectal cancer Jonathan M. Hernandez & Whalen Clark & Jill Weber & William J. Fulp & Lauren Lange & David Shibata
Accepted: 2 June 2014 / Published online: 27 June 2014 # Springer-Verlag Berlin Heidelberg 2014
Abstract Purpose For patients with locally advanced rectal cancer, the accuracy rates of preneoadjuvant therapy nodal staging and potential nodal downstaging make the prognostic significance of nodal status unclear. We therefore sought to review our experience in order to better understand the impact of clinical and pathologic nodal status upon patient outcomes. Methods 174 patients were identified as having undergone neoadjuvant chemoradiation and resection for rectal cancer. For analytic purposes, patients were grouped into four nodal categories (uN 0·pN 0, uN 0·pN +, uN + ·pN 0, and uN + ·pN +). Univariate and multivariate analyses were performed. Results 104 men and 70 women of median age 60 years (29– 85 years) were followed for a median of 31 months (1– 121 months). Nodal staging was available for 129 patients, with a median of 8 lymph nodes (range 0–39) evaluated. Disease recurred in 3 of 41 (7 %) uN 0·pN 0, 10 of 52 (20 %) uN +·pN 0, 7 of 18 (41 %) uN 0·pN +, and 6 of 17 (35 %) uN + ·pN + patients. Th os e pat i en t s ha vi ng no dal downstaging (uN +·pN 0) experienced superior overall survival (p=0.03). Only pathologic nodal status was a significant predictor of both disease-free and overall survival in multivariate modeling. Adjuvant chemotherapy did not impact disease-free or overall survival for patients with pN0. Conclusions Pathologic nodal status may represent a superior predictor of survival for patients with local advanced rectal
J. M. Hernandez : W. Clark : J. Weber : L. Lange : D. Shibata (*) Department of Gastrointestinal Oncology, H. Lee Moffitt Cancer Center and Research Institute, Tampa, FL, USA e-mail: [email protected] W. J. Fulp Biostatistics Core, H. Lee Moffitt Cancer Center and Research Institute, Tampa, FL, USA
cancers. Our findings may have potential implications for the application of adjuvant therapy. Keywords Rectal cancer . Lymph nodes . Neoadjuvant therapy . Resection . EUS
Introduction Rectal cancers extending through the rectal wall or involving locoregional lymph nodes have historically been associated with significant rates of local relapse [1]. The initial application of postoperative [2] and subsequently, preoperative chemoradiation in combination with radical resection has been shown to result in improved outcomes [3]. Consequently, the current preferred standard of care for patients diagnosed with rectal adenocarcinoma and tumor invasion beyond the muscularis propria (T3 or T4) or preoperative nodal involvement (N+) by imaging criteria is neoadjuvant chemoradiation followed by radical resection and subsequent adjuvant chemotherapy [4]. Sole reliance on preoperative imaging modalities for stage assignment has given rise to concerns that a significant number of patients will be overtreated or undertreated. Given that endorectal ultrasound (EUS), CT and pelvic MRI are associated with significant false-negative rates, and that inflammation, scar, stool, and other artifacts can cause false positives during pretreatment scanning [5]; the significance of postoperative pathologic nodal status following rectal resection is of interest. We sought to evaluate whether pathologic nodal stage, which is routinely reported for all patients undergoing resection, can be used as a prognostic variable following multimodality treatment of rectal cancer. We hypothesized that pathologic nodal status would be an important predictor of outcome and may be potentially useful for determining the application of adjuvant therapy in such cases.
1062
Methods Comprehensive clinicopathologic data were collected from 1998 to 2008 for this IRB-approved study. Patients were identified based upon their diagnosis of stage II or III rectal adenocarcinoma and completion of neoadjuvant chemoradiation followed by radical surgical resection. All patients were locally staged prior to receipt of neoadjuvant therapy with endorectal ultrasound and, in some cases, pelvic MRI. Abnormal nodes meeting imaging criteria for positivity were generally not confirmed with biopsy. Neoadjuvant therapy consisted primarily of 5-FU-based chemotherapy, with concurrent external beam radiation (50.4 Gy). Adjuvant chemotherapy consisted of 5-FU, leucovorin, and oxaliplatin (FOLFOX). Disease-free and overall survival analyses were performed using the Kaplan-Meier method and Cox proportional hazard models. Univariate analyses were performed using Chi-squared test, Fisher’s exact test, or ANOVA where appropriate. Significant variables by univariate analyses were considered for multivariate modeling. The final multivariate models were developed by backward elimination using Akaike information criterion. For comparison, patients were grouped into four paired preoperative and postoperative nodal categories (uN 0·pN 0, uN 0·pN +, uN + ·pN 0, and uN + ·pN +).
Int J Colorectal Dis (2014) 29:1061–1068 Table 1 Characteristics of 174 rectal cancer patients
Percent
106 68 60.5
60.9 39.0 12.9
White 165 Nonwhite 7 Unknown 2 Tumor stage (at EUS) T1 1 T2 12 T3 124 T4 6 Nodal stage (at EUS) N0 63 N+ 73 Nx 7 Tumor stage (at resection) T0 42 Tis 1 T1 10 T2 51 T3 64 T4 6 Nodal stage (at resection) N0 117 N1 32 N2 12 Nx 13 Type of surgery LAR 134 APR 34 Other 6
94.8 4.0 1.1
Gender Male Female Age Race
Results We identified 174 patients with locally advanced rectal cancer treated at our institution (104 men and 70 women) with a median age of 60 years (range 29–85). Following completion of neoadjuvant chemoradiation, 140 patients underwent low anterior resection and 34 underwent abdominoperineal resection (APR). Following resection, 90 % of patients completed adjuvant chemotherapy. Comprehensive demographic data is shown in Table 1. Median disease-free and overall survival for the cohort was 66.7 and 73.8 months, respectively. Univariate analyses were undertaken to determine the impact of various conventional parameters upon disease-free and overall survival (Table 2). Interestingly, in our patient population, “T” downstaging had no demonstrable impact upon disease-free or overall survival (Fig. 1a, b). Pretreatment (radiologic) and postoperative (pathologic) nodal staging was available for 129 patients. The median number of lymph nodes evaluated in patients with pretreatment and postoperative nodal staging was 8 (range 0–39).
Number
0.7 8.4 86.7 4.2 44.1 51 4.9
24.1 0.6 5.8 29.3 36.8 3.4 67.2 18.4 6.9 7.5 77 19.5 3.5
metastases on pathologic evaluation. Median disease-free and overall survival for patients with nodal staging of uN 0·pN 0 were not met (Fig. 2a, b). Of the 42 patients staged uN 0·pN 0, 3 (7 %) patients experienced disease recurrence (Table 3). On univariate analysis, patients without lymph node metastases on final pathology had significantly prolonged disease-free survival as compared to patients with lymph node metastases (Fig. 2a, p=0.02).
Nodal status uN 0·pN 0 Forty-two patients did not have evidence of lymph node metastases on preoperative imaging and ultimately no nodal
uN + ·pN 0 Fifty-two patients had evidence of lymph node metastases on preoperative imaging but ultimately did not have nodal metastases on pathologic evaluation. Median disease-free and
Int J Colorectal Dis (2014) 29:1061–1068
1063
Table 2 Univariate model for disease-free and overall survival Variable
Disease-free survival (p value)
Overall survival (p value)
Grade Perineural invasion Margin (circumferential resection margin) Circumferential tumor APR T stage Nodal category EUS nodal stage Pathologic nodal stage Adjuvant chemotherapy
0.23 0.16 0.03
0.22 0.35 0.41
0.77 0.11 0.06 0.003 0.92 0.004 0.90
0.70 0.14 0.26 0.05 0.70 0.05 0.20
uN 0·pN +
1.00
Eighteen patients did not have evidence of lymph node metastases on preoperative imaging but ultimately had nodal
0.50
0.75
T-Stage 0 n=42
0.25
T-Stage 1-4 n=131
p=0.0664
0.00
Probability of Survival
a
0
1
2
3
4
5
T-Stage 0 n=42
0.50
0.75
b
1.00
Time (Yrs)
0.25
T-Stage 1-4 n=131
p=0.3167
0.00
Probability of Survival
Fig. 1 Disease-free (a) and overall survival (b) stratified by tumor stage at resection
overall survival for patients with nodal staging of uN +·pN0 were 73.8 and 85.4 months (Fig. 2a, b), respectively. Of the 52 patients staged as uN +·pN 0, 10 (20 %) patients experienced disease recurrence (Table 3). On univariate analysis, patients without lymph node metastases on final pathology had a significantly prolonged disease-free survival as compared to patients with lymph node metastases (Fig. 2a, p=0.02). Patients experiencing nodal downstaging (uN +·pN 0) experienced superior overall survival as compared to patients experiencing nodal upstaging (uN 0·pN +) (Fig. 2c, p=0.03).
0
1
2 Time (Yrs)
3
4
5
1064
uN0,pN0 (n=42)
0.50
uN+,pN0 (n=52) uN+,pN+ (n=17)
0.25
uN0,pN+ (n=18)
p=0.0222
0.00
Probability of Survival
0.75
1.00
a
b
1
2 Time (Yrs)
3
4
5
1.00
0
0.75 0.25
0.50
uN0,pN0 (n=42) uN+,pN+ (n=17)
uN0,pN+ (n=18)
0.00
Probability of Survival
uN+,pN0 (n=52)
p=0.2167 0
1
2
3
4
5
c
1.00
Time (Yrs)
0.75
uN+,pN0 n=52
uN0,pN+
0.25
0.50
n=18
p=0.03
0.00
Probability of Survival
Fig. 2 Disease-free (a) and overall survival (b, c) stratified by pretherapy (u) and posttherapy (p) nodal groups
Int J Colorectal Dis (2014) 29:1061–1068
0
1
2 Time (Yrs)
3
4
5
Int J Colorectal Dis (2014) 29:1061–1068
1065
Table 3 Locations of first recurrence stratified by nodal group Locations of first recurrence Nodal group uN 0⋅pN 0a uN+⋅pN 0 uN 0⋅pN + a uN+⋅pN+ a
Number 3 10 7 6
Local 1 2 1 2
Liver 1 3 3 2
Lung 1 3 3 1
Other 1 2 3 1
Some patients had synchronous recurrences in more than one location
metastases on pathologic evaluation. Median disease-free and overall survival for patients with nodal staging of uN 0·pN + were 38.1 and 55.7 months (Fig. 2a, b). Of the 18 patients staged uN 0·pN +, 7 (41 %) patients experienced disease recurrence (Table 3). On univariate analysis, patients with lymph node metastases on final pathology had significantly abbreviated disease-free survival as compared to patients without lymph node metastases (Fig. 2a, p=0.02). Patients experiencing nodal upstaging (uN 0·pN +) experienced abbreviated overall survival as compared to patients experiencing nodal downstaging (uN +·pN 0) (Fig. 2c, p=0.03). uN +·pN + Seventeen patients had evidence of lymph node metastases on preoperative imaging and ultimately had persistent nodal metastases on pathologic evaluation. Median disease-free and overall survival for patients with nodal staging of uN +·pN + were 87.1 and 60.5 months (Fig. 2a, b), respectively. Of the 17 patients staged uN +·pN +, 6 (35 %) patients experienced disease recurrence (Table 3). On univariate analysis, patients with lymph node metastases on final pathology had significantly reduced disease-free survival as compared to patients without lymph node metastases (Fig. 2a, p=0.02). Using AIC backward selection, we identified a significant multivariate model to predict disease-free (p=0.0016, Table 4) Table 4 Multivariate model for disease-free survival Variable
Odds ratio
95 % CI
p Value
Age Nodal stage pN0 pN+ Tumor stage T0 T any Type of surgery LAR APR
1.03
0.99–1.05
0.077
– 2.55
– 1.25–5.21
– 0.010
– 2.34
– 0.79–6.90
– 0.12
– 1.89
– 0.87–4.09
– 0.11
and overall survival (p=0.029, Table 5). The variables evaluated in our disease-free and overall survival models included the following: pathologic “T” stage, pathologic “N” stage, age, tumor grade, margin status, circumferential tumor, type of surgery (APR), and adjuvant therapy. Of the individual components of the final disease-free (Table 4) and overall survival (Table 5) models, only pathologic nodal status was a significant prognostic indicator. We also stratified those patients without evidence of lymph node metastases on pathologic examination (uN 0·pN 0 and uN +·pN 0) by receipt of adjuvant chemotherapy. Importantly, adjuvant chemotherapy did not impact disease-free or overall survival (Fig. 3) for patients without evidence of lymph node metastases on final pathology.
Discussion The treatment algorithms for rectal adenocarcinoma are based upon preoperative staging modalities with, at best, a sensitivity of 75 % for nodal metastases with the general assumption that postoperative pathologic evaluation does not necessarily represent the original stage and thus does not accurately provide prognostic information. By default, patients who are generally assigned to neoadjuvant chemoradiation for locally advanced rectal cancer are routinely recommended to also receive adjuvant chemotherapy [4]. Interestingly, for such rectal cancer patients, there are no prospective randomized trials that definitely show a benefit for the application of postoperative adjuvant therapy. Rather, data from colon cancer are extrapolated to this clinical scenario. This approach may potentially augment stage-specific survival given the propensity for false positives, but at what cost? The subset of patients that may truly be node negative and assigned to chemotherapy may unnecessarily suffer the attendant adverse effects of these regimens and may also add to the burden of health care costs. In the present study, we demonstrate that pathologic nodal status is a strong predictor of disease-free and overall survival, irrespective of preoperative stage assignment. Furthermore, we found no benefit from the addition of adjuvant chemotherapy for those patients without evidence of lymphatic Table 5 Multivariate model for overall survival Variable
Odds ratio
95 % CI
p Value
Age Nodal stage pN0 pN+
1.03
0.99–1.06
0.07
– 2.31
– 1.09–4.89
– 0.03
a
Yes
0.25
0.50
No
0.00
Probability of Survival
0.75
Fig. 3 Disease-free (a) and overall survival (b) among patients without evidence of lymph node metastases on pathologic evaluation, stratified by adjuvant chemotherapy
Int J Colorectal Dis (2014) 29:1061–1068
1.00
1066
p=0.9627
0
1
2
3
4
5
b
1.00
Time (Yrs)
0.25
0.50
No
0.00
Probability of Survival
0.75
Yes
p=0.4937
0
1
2
3
4
5
Time (Yrs)
metastases on final pathologic evaluation. Our study suggests that pathologic nodal status may be potentially useful to supplement decision-making with respect to adjuvant treatment. Larger-scale studies of prospective cohorts are certainly warranted. Although the currently available staging modalities are reasonably accurate, they remain imperfect. Based upon the results of large meta-analyses [6], the accuracy of EUS and MRI for determination of T stage is 87 and 71–91 %, respectively. The accuracy of EUS and MRI for determination of “N”
stage is 74 and 45–79 %, respectively. All patients in our study were staged with EUS, and some were additionally imaged with MRI. We recognize the limitations of the current preoperative “T” and “N” staging and the possible tendencies toward overcalling, especially considering that 18 % of nodal metastases occur in lymph nodes less than 5 mm in diameter [7]. Further complicating matters are the lack of accurate predictors of response to neoadjuvant therapy. CEA levels have been suggested in several studies [8–10] as a surrogate marker for response, although cutoff levels have been inconsistent.
Int J Colorectal Dis (2014) 29:1061–1068
Additionally, many studies have attempted to retrospectively correlate immunohistochemical and gene expression biomarkers with response to neoadjuvant therapy [11–18]. At present, the use of biomarkers to predict response remains investigational. Although restaging studies following neoadjuvant therapy are theoretically appealing to help mitigate the inaccuracies of pretreatment stage assignment, none of the currently available examination (DRE, endoscopy) or imaging tools (CT, MR, EUS, PET) are accurate in the setting of fibrosis and inflammation caused by radiation [19]. For example, a prospective study comparing postchemoradiation staging using CT, MR, and EUS to histopathologic findings demonstrated accuracy rates of 37–62, 34–68, and 27–65 %, respectively; most of which was due to overstaging [20]. We do not routinely locally restage patients following neoadjuvant therapy for the reasons stated above, although physical and endoscopic examination prior to resection is generally advised. We did not find any correlation between primary tumor (T) downstaging and survival endpoints. In comparing total primary tumor regression (pT0) to any residual primary tumor (pT+), we did not find any significant correlation with diseasefree or overall survival. Similar results were demonstrated in a multivariate model constructed on a cohort of 153 patients with advanced rectal cancer, in which tumor regression grade is not associated with survival endpoints [21]. Interestingly, other studies have demonstrated a correlation between poor response to neoadjuvant therapy and increased rates of local recurrence [22]. Perhaps, these disparate findings may be the result of ill-defined points of optimal cutoff for separating good and poor response, as has been recently suggested [23]. This problem has been recently addressed by Dhadda et al. [24], who developed a scoring system incorporating cut point tumor regression grades (TRGs), although large-scale validation will be required. Although there is no minimum lymph node retrieval suggested for rectal cancer operations (as opposed to colon cancer), a mean of eight lymph nodes was evaluated for the patients in our study, which is comparable to other series with similar patients [25, 26]. Interestingly, retrieval of
