ORIGINAL ARTICLE
Staging of early lymph node metastases with the sentinel lymph node technique and predictive factors in T1/T2 oral cavity cancer: A retrospective single-center study Nicklas Juel Pedersen, BSc, Med,1 David Hebbelstrup Jensen, MD,1 Nora Hedb€ack, MD,1 Martin Frendø, BSc, Med,1 Katalin Kiss, MD,2 Giedrius Lelkaitis, MD,2 Jann Mortensen, MD, DMSc,3 Anders Christensen, MD,1 Lena Specht, MD, DMSc,4 Christian von Buchwald, MD, DMSc1* 1
Department of Otorhinolaryngology, Head and Neck Surgery and Audiology, Rigshospitalet, University of Copenhagen, Copenhagen, Denmark, 2Department of Pathology, Rigshospitalet, University of Copenhagen, Denmark, 3Department of Clinical Physiology, Nuclear Medicine and PET, Rigshospitalet, University of Copenhagen, Copenhagen, Denmark, 4 Department of Oncology, Rigshospitalet, University of Copenhagen, Copenhagen, Denmark.
Accepted 31 May 2015 Published online 2 September 2015 in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/hed.24153
ABSTRACT: Background. The purpose of this study was to examine the diagnostic accuracy of detecting lymph node metastases and to identify predictive and prognostic clinicopathological factors in patients with oral squamous cell carcinoma (OSCC) undergoing sentinel lymph node biopsy (SLNB). Methods. All patients diagnosed with cT1 to T2N0 OSCC who underwent a diagnostic SLNB between 2007 and 2013 were included. Results. We identified 253 patients, of whom 27% had a positive sentinel lymph node (SLB). The false-negative rate, sensitivity, and negative predictive value (NPV) were 5%, 88%, and 95%, respectively. Patients with micrometastases as well as macrometastases had a separately,
significantly shorter disease-specific survival than patients with pN0 disease. In a logistic regression model, the maximum tumor thickness, perineural invasion, and differentiation grade were independent predictive factors for the presence of metastases. Conclusion. These data support the use of the SLNB technique as an accurate and safe staging tool in patients with OSCC with a cN0 neck. C 2015 Wiley Periodicals, Inc. Head Neck 38: E1033–E1040, 2016 V
INTRODUCTION
biopsy (SLNB) has been developed and advocated. The technique is widely used and carries less morbidity than a full neck dissection,8 but it has not yet been accepted as a standard of care in all cancer centers. The SLNB technique enables detection of micrometastases by step-serial sectioning of the sentinel lymph nodes (SLNs) and is thus labor intensive. A recent metaanalysis has found a sensitivity when using this technique for detecting lymph node metastases of 93% (95% confidence interval [CI] 5 90–95) and a negative predictive value (NPV) between 88% and 100%.9 Since April 2007, the SLNB technique has been implemented as a standard of care in the diagnostic workup of patients with cN0 OSCC at Rigshospitalet, Denmark. The purpose of this study was to examine the reliability of the SLNB technique in daily clinical practice and to examine under which circumstances challenges to the technique are likely to appear. Furthermore, we wanted to identify predictive and prognostic histopathological and clinical parameters for patients with early cT1 to T2N0 OSCC in order to potentially optimize the diagnostic workup as well as treatment planning and follow-up in patients with early OSCC.
Identifying small lymph node metastases in patients with early oral squamous cell carcinoma (OSCC) using traditional imaging (CT, positron emission tomography–CT, and/ or MRI) is known to be unreliable.1,2 Thus, it has been estimated that approximately 20% to 30% of patients with a clinically N0 neck (cN0) will have occult neck disease after careful pathological examination of lymph nodes from an elective neck dissection (pN-positive).3–5 An often cited dogma in the treatment of the cN0 neck in patients with OSCC states that if there is more than 20% risk of lymph node metastases, then elective neck dissection is warranted.6 Historically, the cN0 patient group has been offered upfront neck dissection as opposed to treatment of neck recurrences, because this will increase the chance of long-term survival.7 However, this leads to a dilemma, because between 70% and 80% of the patients will receive overtreatment (ie, neck dissection). Therefore, a better staging of patients with occult lymph node metastases with sentinel lymph node
*Corresponding author: C. von Buchwald, Department of Otorhinolaryngology, Head and Neck Surgery and Audiology, Rigshospitalet, F2073, Blegdamsvej 9, 2100 Copenhagen, Denmark. E-mail: [email protected] Contract grant sponsor: N.J.P. received a grant from the Danish Council for Independent Research (ID: DFF – 4004-00193B) and D.H.J. received a grant from the nonprofit Candy Foundation.
KEY WORDS: sentinel lymph node biopsy, oral squamous cell carcinoma, negative predictive value, survival, predictive histopathological factors
MATERIALS AND METHODS Patients Patients were identified by searching the Danish Pathology Databank specifically for patients treated at Rigshospitalet, HEAD & NECK—DOI 10.1002/HED
APRIL 2016
E1033
PEDERSEN ET AL.
FIGURE 1. Flowchart showing the reasons for exclusion with the distribution of recurrences at the N site and treatment. *Adjuvant radiotherapy to the N site. †Compared to the levels previously biopsied during the SLNB. ‡Three patients received adjuvant radiotherapy without neck dissection (RT) in combination with cisplatin and 1 patient received RT in combination with cisplatin and nimorazole. Nine of the 19 patients treated with RT alone (47%) had extracapsular invasion and 2 of these patients had recurrences at the N site during follow-up. The figure was made at www.draw.io. HNSCC, head and neck squamous cell carcinoma; OSCC, oral squamous cell carcinoma; Neck, neck dissection; PRT, adjuvant radiotherapy to the N site after a neck dissection; WW, watchful waiting; DFT, declined further treatment.
Copenhagen, Denmark, using oral-specific SNOMED codes (T51, T53, and T54) and a code for the specific center.10 The Danish Pathology Databank is a nationwide databank that contains information on all pathological examinations performed in Denmark, which are reported by pathology departments through an online, real-time system.11 We found 1321 potentially eligible patients who were coded with oral-specific SNOMED codes at our center from April 2007 through December 2013, of whom 1068 were excluded (see the details for exclusion in Figure 1). All patients had cT1 to T2N0 OSCC. Patients were classified as cN0 after having undergone clinical examination, endoscopic evaluation of the upper airways, and CT and/or MRI of the oral cavity and neck. In addition, according to international guidelines, patients who had previous surgery or radiotherapy (RT) at the T site and N site were excluded because of possible altered lymph drainage.12 A small fraction of the patients in this study cohort are included in the Sentinel European Node Trial,13 and have previously been included in other studies by our group.14,15
4 peritumoral submucosal injection sites surrounding the tumor with a dose of 0.2 mL each while the patient is under local anesthesia with lidocaine spray (10 mg/dose). No blue dye or marking of the skin was used to identify the SLN. For same-day procedures, a median dose of 57 MBq (range, 36–118 MBq) was injected, whereas a median dose of 114 MBq (range, 55–124 MBq) was injected for 2-day procedures. All patients underwent dynamic and static lymphoscintigraphy with single photon emission CT (SPECT/CT) according to joint sentinel lymph node guidelines either the day before surgery or on the morning of the day of surgery.12 During surgery, tumor resection was performed before the neck procedure to reduce perioperative radioactive disturbance when using the handheld g-probe to localize the SLN(s).12 Skip metastases were defined as nodal spread to levels IV to VI without involvement of levels I to III. Follow-up of the SLNB-negative patients was used as the reference test for the sensitivity of the SLNB technique.
Sentinel lymph node biopsy technique
Histopathological examination
The Tc-99-labeled human serum nanocolloid (Nanocoll; GE Healthcare, Brøndby, Denmark) tracer was injected at
The SLNs were histopathologically examined according to the Canniesburn Protocol.16 We used monoclonal
E1034
HEAD & NECK—DOI 10.1002/HED
APRIL 2016
SENTINEL
LYMPH NODE BIOPSY IN ORAL CANCER
FIGURE 2. Patients with micrometastases and macrometastases have a shorter survival based on Kaplan–Meier plots. (A) Disease-free survival, (B) disease-specific survival. Equalities of the survival distributions were calculated by a Mantel–Cox log-rank test.
mouse anti-human AE1/AE3 antibody (DAKO, Glostrup, Denmark) and the OptiView detection kit on a Ventana BenchMark ULTRA autostainer (Roche, Lyngby, Denmark). Occult metastases were histopathologically divided into isolated tumor cells (ITC), (2 mm).17 We defined a histologically free margin of the primary tumor resection specimen to be >5 mm, close margins to be between 1 and 5 mm, and involved margins to be pN1 disease, or nonradical excision at the T site) or the patients declined further surgery, RT was offered. At our center, patients are followed for a standard period of 5 years, consistent with Danish national guidelines.19
We collected information on tumor location, surgical treatment, recurrences, survival (including causes of death from hospitals records), adjuvant therapy, and status at last follow-up. Alcohol consumption was defined as high if the weekly intake exceeded 87.5 g of alcohol for women and 175 g for men, in accordance with the Danish health recommendation.24 Tobacco consumption was defined as high if the patient reported a history of >10 pack-years and as low if it was 10 pack-years. The study was conducted in accordance with the Helsinki Declaration and approved by the Danish Regional Scientific Ethical Committee (H-1-2014-H53) and the Danish Data Protection Agency.
Statistics Statistical analyses were performed in SPSS version 21 (SPSS, Chicago, IL). Survival curves were analyzed by the Kaplan–Meier method and the corresponding p values were calculated by Mantel–Cox log-rank tests.25 Endpoints in the survival analyses were disease-free survival (DFS) and disease-specific survival (DSS). The DSS was defined as the time from diagnosis of malignancy to death caused by OSCC, whereas deaths not caused by OSCC were censored at the last known follow-up. The DFS was defined as the time from the diagnosis of malignancy to recurrence at the T, N, or M site. Deaths from any cause were censored at the last known follow-up. Hazard ratios were calculated by univariate Cox regression analyses for each parameter (Table 1). We calculated the area under the curve (AUC) of the receiver operating characteristic curves to examine the ability of various parameters to discriminate between pN-positive and pN-negative (Table 2). To combine the predictive factors for lymph node metastases, we used predicted probabilities from binary logistic regression with backward elimination to compute the AUC. A predictive model was made from the binary logistic regression of the independently predictive histopathological parameters to calculate the risk of occult lymph node HEAD & NECK—DOI 10.1002/HED
APRIL 2016
E1035
PEDERSEN ET AL.
TABLE 1. Clinical and pathological characteristics in relation to disease-free and disease-specific survival by univariate analyses. DFS – 83% Variables
Metastases No spread Micrometastases and isolated tumor cells Macrometastases Macrometastases vs micrometastases and isolated tumor cells Extracapsular spread cT2/T1 Maximum tumor thickness* Tumor invasive front† Perineural invasion Vascular invasion p16 expression Differentiation grade Well Moderate Poor Tumor location FOM Tongue Other tumor subsites‡ Age, per 1 y increments Sex, male/female Smoking, high/low Alcohol intake, high/low
No. of events
HR (95% CI)
43 22 6
DSS – 93% p value
No. of events
HR (95% CI)
p value
1 2.25 (0.91–5.54)
1 .08
17 5 4
1 6.26 (1.68–23.3)
1 < .01
15 21
3.42 (1.77–6.60) 1.52 (0.59–3.92)
< .01 .39
8 12
8.18 (2.67–25.10) 1.32 (0.40–4.41)
< .01 .65
43 43 43 40 40 41 29 43 8 25 10 43 21 16 6 43 43 38 43
1.66 (0.51–5.35) 2.43 (1.34–4.43) 1.12 (1.03–1.22) 2.30 (1.14–4.62) 2.64 (1.41–4.95) 1.30 (0.46–3.64) 0.59 (0.18–1.94)
.40 < .01 .01 .02 < .01 .62 .38
1.32 (0.18–9.96) 3.30 (1.26–8.67) 1.21 (1.07–1.37) 2.20 (0.77–6.27) 3.49 (1.34–9.08) 1.55 (0.35–6.78) 0.41 (0.05–3.18)
.79 .02 < .01 .14 .01 .56 .40
1 3.06 (1.21–7.75) 1.71 (0.77–3.81)
1 .02 .19
1 1.51 (0.48–4.74) 1.10 (0.20–6.05)
1 .49 .91
1 0.87 (0.46–1.68) 1.41 (0.57–3.49) 1.01 (0.98–1.03) 0.64 (0.35–1.17) 0.75 (0.36–1.55) 1.18 (0.64–2.17)
1 .69 .46 .59 .15 .44 .59
17 17 17 17 17 17 13 17 4 11 2 17 5 9 3 17 17 14 17
1 2.13 (0.71–6.40) 2.77 (0.66–11.63) 1.02 (0.98–1.06) 0.86 (0.33–2.24) 0.54 (0.18–1.62) 0.80 (0.31–2.01)
1 .18 .16 .44 .76 .27 .66
Abbreviations: DFS, disease-free survival; DSS, disease-specific survival; HR, hazard ratio; CI, confidence interval; FOM, floor of mouth. * The maximum tumor thickness was used as a continuous variable (ie, no specific cutoff). † Noncohesive vs cohesive invasive tumor front. ‡ Oral squamous cell carcinoma of the buccal mucosa, gingiva, and retromolar sites were pooled because of the low number of events.
metastases using only T site parameters. Significance was set at p .05.
Missing data Because of insufficient access or missing data, we could not obtain information on all clinical and pathological parameters. Missing data were left out of the analysis. In cases of missing TNM classifications, the T classification was defined from the size measured at the preoperative physical examination.
RESULTS During the study period from April 2007 through December 2013, 253 consecutive patients with cT1 to T2 tumors who were clinically and radiologically confirmed not to have lymph node metastases (cN0) underwent SLNB. The median follow-up was 32 months (range, 1– 92 months). At the time of diagnosis, the patients had a median age of 63 years (range, 30–95 years) and 145 were men (57%). The preoperative SPECT/CT lymphoscintigraphy identified a median of 2 SLNs (range, 0–9) whereas a median of 3 SLNs were harvested during the SLNB procedure (range, 1–12). In 1 patient, the SPECT/CT was inconclusive because of comorbidities, but the perioperative g-probe detected the SLNs successfully. The SPECT/CT E1036
HEAD & NECK—DOI 10.1002/HED
APRIL 2016
lymphoscintigraphy demonstrated bilateral drainage in 32% of the patients (80 of 253), of which tumors from the floor of the mouth (FOM) had bilateral drainage in 43% of the patients (52 of 120). No severe adverse reactions to the tracer substances were reported.
Detection of occult lymph node metastases in patients with cN0 oral squamous cell carcinoma Of the 253 SLNB patients, 27% (n 5 68) were found to harbor occult lymph node metastases. Of the 68 patients with occult lymph node disease, 29% had more than 1 positive lymph nodes (20 of 68), and in 56% of the cases, the metastasis was found in the hottest lymph node (highest radioactive count, 38 of 68). Further examination of the 88 positive SLNs revealed that 2 patients only had ITC, whereas 21 harbored micrometastases and 45 harbored macrometastases. ECS was observed in 21% of the SLNB-positive patients (14 of 68, of whom 2 patients with micrometastases had ECS). Skip metastases were identified in level IV in 2 patients with tongue tumors (of whom 1 patient had a tumor close to the apex of the tongue) and in 1 patient with a FOM tumor. All patients with skip metastases subsequently underwent neck dissection of levels II to V and no additional metastases were found. An unexpected pattern of metastatic spread to level IIB was observed in 2 patients;
SENTINEL
LYMPH NODE BIOPSY IN ORAL CANCER
TABLE 2. Predictive factors for lymph node involvement.
Tumor characteristics
Overall Clinical T classification T2 T1 Grade of differentiation Well Moderate Poor Invasive front Cohesive Noncohesive Perineural invasion Yes No Vascular invasion Yes No p16 expression Positive Negative Tumor invasion depth
Staging of early lymph node metastases with the sentinel lymph node technique and predictive factors in T1/T2 oral cavity cancer: A retrospective single-center study Nicklas Juel Pedersen, BSc, Med,1 David Hebbelstrup Jensen, MD,1 Nora Hedb€ack, MD,1 Martin Frendø, BSc, Med,1 Katalin Kiss, MD,2 Giedrius Lelkaitis, MD,2 Jann Mortensen, MD, DMSc,3 Anders Christensen, MD,1 Lena Specht, MD, DMSc,4 Christian von Buchwald, MD, DMSc1* 1
Department of Otorhinolaryngology, Head and Neck Surgery and Audiology, Rigshospitalet, University of Copenhagen, Copenhagen, Denmark, 2Department of Pathology, Rigshospitalet, University of Copenhagen, Denmark, 3Department of Clinical Physiology, Nuclear Medicine and PET, Rigshospitalet, University of Copenhagen, Copenhagen, Denmark, 4 Department of Oncology, Rigshospitalet, University of Copenhagen, Copenhagen, Denmark.
Accepted 31 May 2015 Published online 2 September 2015 in Wiley Online Library (wileyonlinelibrary.com). DOI 10.1002/hed.24153
ABSTRACT: Background. The purpose of this study was to examine the diagnostic accuracy of detecting lymph node metastases and to identify predictive and prognostic clinicopathological factors in patients with oral squamous cell carcinoma (OSCC) undergoing sentinel lymph node biopsy (SLNB). Methods. All patients diagnosed with cT1 to T2N0 OSCC who underwent a diagnostic SLNB between 2007 and 2013 were included. Results. We identified 253 patients, of whom 27% had a positive sentinel lymph node (SLB). The false-negative rate, sensitivity, and negative predictive value (NPV) were 5%, 88%, and 95%, respectively. Patients with micrometastases as well as macrometastases had a separately,
significantly shorter disease-specific survival than patients with pN0 disease. In a logistic regression model, the maximum tumor thickness, perineural invasion, and differentiation grade were independent predictive factors for the presence of metastases. Conclusion. These data support the use of the SLNB technique as an accurate and safe staging tool in patients with OSCC with a cN0 neck. C 2015 Wiley Periodicals, Inc. Head Neck 38: E1033–E1040, 2016 V
INTRODUCTION
biopsy (SLNB) has been developed and advocated. The technique is widely used and carries less morbidity than a full neck dissection,8 but it has not yet been accepted as a standard of care in all cancer centers. The SLNB technique enables detection of micrometastases by step-serial sectioning of the sentinel lymph nodes (SLNs) and is thus labor intensive. A recent metaanalysis has found a sensitivity when using this technique for detecting lymph node metastases of 93% (95% confidence interval [CI] 5 90–95) and a negative predictive value (NPV) between 88% and 100%.9 Since April 2007, the SLNB technique has been implemented as a standard of care in the diagnostic workup of patients with cN0 OSCC at Rigshospitalet, Denmark. The purpose of this study was to examine the reliability of the SLNB technique in daily clinical practice and to examine under which circumstances challenges to the technique are likely to appear. Furthermore, we wanted to identify predictive and prognostic histopathological and clinical parameters for patients with early cT1 to T2N0 OSCC in order to potentially optimize the diagnostic workup as well as treatment planning and follow-up in patients with early OSCC.
Identifying small lymph node metastases in patients with early oral squamous cell carcinoma (OSCC) using traditional imaging (CT, positron emission tomography–CT, and/ or MRI) is known to be unreliable.1,2 Thus, it has been estimated that approximately 20% to 30% of patients with a clinically N0 neck (cN0) will have occult neck disease after careful pathological examination of lymph nodes from an elective neck dissection (pN-positive).3–5 An often cited dogma in the treatment of the cN0 neck in patients with OSCC states that if there is more than 20% risk of lymph node metastases, then elective neck dissection is warranted.6 Historically, the cN0 patient group has been offered upfront neck dissection as opposed to treatment of neck recurrences, because this will increase the chance of long-term survival.7 However, this leads to a dilemma, because between 70% and 80% of the patients will receive overtreatment (ie, neck dissection). Therefore, a better staging of patients with occult lymph node metastases with sentinel lymph node
*Corresponding author: C. von Buchwald, Department of Otorhinolaryngology, Head and Neck Surgery and Audiology, Rigshospitalet, F2073, Blegdamsvej 9, 2100 Copenhagen, Denmark. E-mail: [email protected] Contract grant sponsor: N.J.P. received a grant from the Danish Council for Independent Research (ID: DFF – 4004-00193B) and D.H.J. received a grant from the nonprofit Candy Foundation.
KEY WORDS: sentinel lymph node biopsy, oral squamous cell carcinoma, negative predictive value, survival, predictive histopathological factors
MATERIALS AND METHODS Patients Patients were identified by searching the Danish Pathology Databank specifically for patients treated at Rigshospitalet, HEAD & NECK—DOI 10.1002/HED
APRIL 2016
E1033
PEDERSEN ET AL.
FIGURE 1. Flowchart showing the reasons for exclusion with the distribution of recurrences at the N site and treatment. *Adjuvant radiotherapy to the N site. †Compared to the levels previously biopsied during the SLNB. ‡Three patients received adjuvant radiotherapy without neck dissection (RT) in combination with cisplatin and 1 patient received RT in combination with cisplatin and nimorazole. Nine of the 19 patients treated with RT alone (47%) had extracapsular invasion and 2 of these patients had recurrences at the N site during follow-up. The figure was made at www.draw.io. HNSCC, head and neck squamous cell carcinoma; OSCC, oral squamous cell carcinoma; Neck, neck dissection; PRT, adjuvant radiotherapy to the N site after a neck dissection; WW, watchful waiting; DFT, declined further treatment.
Copenhagen, Denmark, using oral-specific SNOMED codes (T51, T53, and T54) and a code for the specific center.10 The Danish Pathology Databank is a nationwide databank that contains information on all pathological examinations performed in Denmark, which are reported by pathology departments through an online, real-time system.11 We found 1321 potentially eligible patients who were coded with oral-specific SNOMED codes at our center from April 2007 through December 2013, of whom 1068 were excluded (see the details for exclusion in Figure 1). All patients had cT1 to T2N0 OSCC. Patients were classified as cN0 after having undergone clinical examination, endoscopic evaluation of the upper airways, and CT and/or MRI of the oral cavity and neck. In addition, according to international guidelines, patients who had previous surgery or radiotherapy (RT) at the T site and N site were excluded because of possible altered lymph drainage.12 A small fraction of the patients in this study cohort are included in the Sentinel European Node Trial,13 and have previously been included in other studies by our group.14,15
4 peritumoral submucosal injection sites surrounding the tumor with a dose of 0.2 mL each while the patient is under local anesthesia with lidocaine spray (10 mg/dose). No blue dye or marking of the skin was used to identify the SLN. For same-day procedures, a median dose of 57 MBq (range, 36–118 MBq) was injected, whereas a median dose of 114 MBq (range, 55–124 MBq) was injected for 2-day procedures. All patients underwent dynamic and static lymphoscintigraphy with single photon emission CT (SPECT/CT) according to joint sentinel lymph node guidelines either the day before surgery or on the morning of the day of surgery.12 During surgery, tumor resection was performed before the neck procedure to reduce perioperative radioactive disturbance when using the handheld g-probe to localize the SLN(s).12 Skip metastases were defined as nodal spread to levels IV to VI without involvement of levels I to III. Follow-up of the SLNB-negative patients was used as the reference test for the sensitivity of the SLNB technique.
Sentinel lymph node biopsy technique
Histopathological examination
The Tc-99-labeled human serum nanocolloid (Nanocoll; GE Healthcare, Brøndby, Denmark) tracer was injected at
The SLNs were histopathologically examined according to the Canniesburn Protocol.16 We used monoclonal
E1034
HEAD & NECK—DOI 10.1002/HED
APRIL 2016
SENTINEL
LYMPH NODE BIOPSY IN ORAL CANCER
FIGURE 2. Patients with micrometastases and macrometastases have a shorter survival based on Kaplan–Meier plots. (A) Disease-free survival, (B) disease-specific survival. Equalities of the survival distributions were calculated by a Mantel–Cox log-rank test.
mouse anti-human AE1/AE3 antibody (DAKO, Glostrup, Denmark) and the OptiView detection kit on a Ventana BenchMark ULTRA autostainer (Roche, Lyngby, Denmark). Occult metastases were histopathologically divided into isolated tumor cells (ITC), (2 mm).17 We defined a histologically free margin of the primary tumor resection specimen to be >5 mm, close margins to be between 1 and 5 mm, and involved margins to be pN1 disease, or nonradical excision at the T site) or the patients declined further surgery, RT was offered. At our center, patients are followed for a standard period of 5 years, consistent with Danish national guidelines.19
We collected information on tumor location, surgical treatment, recurrences, survival (including causes of death from hospitals records), adjuvant therapy, and status at last follow-up. Alcohol consumption was defined as high if the weekly intake exceeded 87.5 g of alcohol for women and 175 g for men, in accordance with the Danish health recommendation.24 Tobacco consumption was defined as high if the patient reported a history of >10 pack-years and as low if it was 10 pack-years. The study was conducted in accordance with the Helsinki Declaration and approved by the Danish Regional Scientific Ethical Committee (H-1-2014-H53) and the Danish Data Protection Agency.
Statistics Statistical analyses were performed in SPSS version 21 (SPSS, Chicago, IL). Survival curves were analyzed by the Kaplan–Meier method and the corresponding p values were calculated by Mantel–Cox log-rank tests.25 Endpoints in the survival analyses were disease-free survival (DFS) and disease-specific survival (DSS). The DSS was defined as the time from diagnosis of malignancy to death caused by OSCC, whereas deaths not caused by OSCC were censored at the last known follow-up. The DFS was defined as the time from the diagnosis of malignancy to recurrence at the T, N, or M site. Deaths from any cause were censored at the last known follow-up. Hazard ratios were calculated by univariate Cox regression analyses for each parameter (Table 1). We calculated the area under the curve (AUC) of the receiver operating characteristic curves to examine the ability of various parameters to discriminate between pN-positive and pN-negative (Table 2). To combine the predictive factors for lymph node metastases, we used predicted probabilities from binary logistic regression with backward elimination to compute the AUC. A predictive model was made from the binary logistic regression of the independently predictive histopathological parameters to calculate the risk of occult lymph node HEAD & NECK—DOI 10.1002/HED
APRIL 2016
E1035
PEDERSEN ET AL.
TABLE 1. Clinical and pathological characteristics in relation to disease-free and disease-specific survival by univariate analyses. DFS – 83% Variables
Metastases No spread Micrometastases and isolated tumor cells Macrometastases Macrometastases vs micrometastases and isolated tumor cells Extracapsular spread cT2/T1 Maximum tumor thickness* Tumor invasive front† Perineural invasion Vascular invasion p16 expression Differentiation grade Well Moderate Poor Tumor location FOM Tongue Other tumor subsites‡ Age, per 1 y increments Sex, male/female Smoking, high/low Alcohol intake, high/low
No. of events
HR (95% CI)
43 22 6
DSS – 93% p value
No. of events
HR (95% CI)
p value
1 2.25 (0.91–5.54)
1 .08
17 5 4
1 6.26 (1.68–23.3)
1 < .01
15 21
3.42 (1.77–6.60) 1.52 (0.59–3.92)
< .01 .39
8 12
8.18 (2.67–25.10) 1.32 (0.40–4.41)
< .01 .65
43 43 43 40 40 41 29 43 8 25 10 43 21 16 6 43 43 38 43
1.66 (0.51–5.35) 2.43 (1.34–4.43) 1.12 (1.03–1.22) 2.30 (1.14–4.62) 2.64 (1.41–4.95) 1.30 (0.46–3.64) 0.59 (0.18–1.94)
.40 < .01 .01 .02 < .01 .62 .38
1.32 (0.18–9.96) 3.30 (1.26–8.67) 1.21 (1.07–1.37) 2.20 (0.77–6.27) 3.49 (1.34–9.08) 1.55 (0.35–6.78) 0.41 (0.05–3.18)
.79 .02 < .01 .14 .01 .56 .40
1 3.06 (1.21–7.75) 1.71 (0.77–3.81)
1 .02 .19
1 1.51 (0.48–4.74) 1.10 (0.20–6.05)
1 .49 .91
1 0.87 (0.46–1.68) 1.41 (0.57–3.49) 1.01 (0.98–1.03) 0.64 (0.35–1.17) 0.75 (0.36–1.55) 1.18 (0.64–2.17)
1 .69 .46 .59 .15 .44 .59
17 17 17 17 17 17 13 17 4 11 2 17 5 9 3 17 17 14 17
1 2.13 (0.71–6.40) 2.77 (0.66–11.63) 1.02 (0.98–1.06) 0.86 (0.33–2.24) 0.54 (0.18–1.62) 0.80 (0.31–2.01)
1 .18 .16 .44 .76 .27 .66
Abbreviations: DFS, disease-free survival; DSS, disease-specific survival; HR, hazard ratio; CI, confidence interval; FOM, floor of mouth. * The maximum tumor thickness was used as a continuous variable (ie, no specific cutoff). † Noncohesive vs cohesive invasive tumor front. ‡ Oral squamous cell carcinoma of the buccal mucosa, gingiva, and retromolar sites were pooled because of the low number of events.
metastases using only T site parameters. Significance was set at p .05.
Missing data Because of insufficient access or missing data, we could not obtain information on all clinical and pathological parameters. Missing data were left out of the analysis. In cases of missing TNM classifications, the T classification was defined from the size measured at the preoperative physical examination.
RESULTS During the study period from April 2007 through December 2013, 253 consecutive patients with cT1 to T2 tumors who were clinically and radiologically confirmed not to have lymph node metastases (cN0) underwent SLNB. The median follow-up was 32 months (range, 1– 92 months). At the time of diagnosis, the patients had a median age of 63 years (range, 30–95 years) and 145 were men (57%). The preoperative SPECT/CT lymphoscintigraphy identified a median of 2 SLNs (range, 0–9) whereas a median of 3 SLNs were harvested during the SLNB procedure (range, 1–12). In 1 patient, the SPECT/CT was inconclusive because of comorbidities, but the perioperative g-probe detected the SLNs successfully. The SPECT/CT E1036
HEAD & NECK—DOI 10.1002/HED
APRIL 2016
lymphoscintigraphy demonstrated bilateral drainage in 32% of the patients (80 of 253), of which tumors from the floor of the mouth (FOM) had bilateral drainage in 43% of the patients (52 of 120). No severe adverse reactions to the tracer substances were reported.
Detection of occult lymph node metastases in patients with cN0 oral squamous cell carcinoma Of the 253 SLNB patients, 27% (n 5 68) were found to harbor occult lymph node metastases. Of the 68 patients with occult lymph node disease, 29% had more than 1 positive lymph nodes (20 of 68), and in 56% of the cases, the metastasis was found in the hottest lymph node (highest radioactive count, 38 of 68). Further examination of the 88 positive SLNs revealed that 2 patients only had ITC, whereas 21 harbored micrometastases and 45 harbored macrometastases. ECS was observed in 21% of the SLNB-positive patients (14 of 68, of whom 2 patients with micrometastases had ECS). Skip metastases were identified in level IV in 2 patients with tongue tumors (of whom 1 patient had a tumor close to the apex of the tongue) and in 1 patient with a FOM tumor. All patients with skip metastases subsequently underwent neck dissection of levels II to V and no additional metastases were found. An unexpected pattern of metastatic spread to level IIB was observed in 2 patients;
SENTINEL
LYMPH NODE BIOPSY IN ORAL CANCER
TABLE 2. Predictive factors for lymph node involvement.
Tumor characteristics
Overall Clinical T classification T2 T1 Grade of differentiation Well Moderate Poor Invasive front Cohesive Noncohesive Perineural invasion Yes No Vascular invasion Yes No p16 expression Positive Negative Tumor invasion depth
