Clinical Nodal Stage Is an Independently Significant Predictor of Distant Failure in Patients With Squamous Cell Carcinoma of the Larynx

Abstract

Objective:

To determine the impact of clinical nodal stage on distant metastasis (DM) in patients with squamous cell carcinoma of the larynx (SCCL).

Methods:

Six hundred sixty-two previously untreated SCCL patients treated at a tertiary care cancer center from January 1984 to December 1998 were eligible for analysis. The end point of interest was development of DM following treatment. Distant metastasis-free survival (DMFS) was calculated by the Kaplan-Meier method; predictors of outcome were identified by univariate and multivariate analysis. The primary tumor site was glottic in 55%, supraglottic in 40%, and trans/sub glottic in 5%; 40% had locoregionally advanced (stage III/IV) tumors. At initial presentation, 25% of patients (12% N1, 11% N2, and 2% N3) had clinically metastatic nodes.

Results:

DM were recorded in 67 patients (10%; lung, 45%; soft tissue, 13%; bone, 10%; multiple sites, 28%). The median time to DM was 18 months (range, 1-109). With a median follow-up of 60 months, the 5-year DMFS was 88%. Even after accounting for the type of index treatment, the only significant predictor of worse DMFS on multivariate analysis was a higher clinical N stage (P < 0.0001). The relative risk for DM was 0.5 (95% CI, 0.2-1.4; P = NS) for cN1, 3.2 (95% CI, 1.7-5.9; P < 0.0001) for cN2, and 7.5 (95% CI, 3.1-17.9; P < 0.0001) for cN3 disease compared with clinically N0 patients.

Conclusion:

Regardless of the index treatment modality, primary tumor site, or T stage, a higher clinical N stage at the time of presentation independently and significantly increases the risk of DM in patients with SCCL.

A large cohort of patients treated for squamous carcinoma of the larynx was analyzed for the relationship of neck node status to distant metastasis. Regardless of treatment modality, tumor site, or T stage, a higher clinical N stage at the time of presentation independently and significantly increased the risk of distant metastasis.

The phenomenon of distant metastasis (DM) after therapy for squamous cell carcinoma of the head and neck (SCCHN) is well recognized, and it has indeed been held responsible for the fact that survival rates for most advanced head and neck cancers have remained constant in spite of improved locoregional control rates in recent years.¹ The need for developing more effective systemic treatment strategies is therefore obvious, but it is equally important to be able to selectively target patients who are at high risk for developing distant failure. It is well recognized that tumors at certain locations such as the hypopharynx have a higher propensity to spread distantly than those at other sites such as the lips and the oral cavity.^2,3 The larynx occupies an intermediate position in this risk stratification, and it is generally true that within the larynx, glottic tumors are less prone to DM than those at other subsites. The most often quoted reason for these observations is that the risk of DM is directly dependent on the incidence of regional nodal metastasis from any given primary site. While there are reports alluding to this relationship, the question has not been examined systematically in a large cohort of patients with SCCL. To the best of our knowledge, the risk of developing DM has never been quantified relative to the volume of clinical nodal disease at presentation in a comprehensive population of patients undergoing multidisciplinary treatment of SCCL. We have recently analyzed a large cohort with SCCL treated at a tertiary care cancer center using a multidisciplinary approach, and this report aims to examine the relationship between the extent of clinical neck node metastasis at presentation and the likelihood of developing DM following contemporary modalities of treatment.

PATIENTS AND METHODS

Seven hundred nineteen patients with a diagnosis of squamous cell carcinoma of the larynx (SCCL) registered between January 1984 and December 1998 at the Memorial Sloan-Kettering Cancer Center (MSKCC) were identified from prospectively maintained institutional databases. Patients who were seen in consultation only (n = 36; 5%), those who did not start/complete their planned therapy at MSKCC (n = 11; 1.5%), those who had carcinoma in situ (n = 7; 1%), and those whose clinical details were not available (n = 3; 0.4%) were excluded from the study. The remaining 662 patients were eligible for this analysis.

A detailed review of the clinical records of eligible patients was performed. Patient demographics and tumor-specific data were entered into a computer database (Microsoft Access 2000; Microsoft Corporation, Redmond, WA), and statistical analyses were performed using a commercially available computer software package (JMP 4.0; SAS Institute Inc., Cary, NC). The primary end point of interest for this study was development of DM following treatment. The DM-free interval was calculated in months from the date of index treatment to the date of clinical or radiologic documentation of DM. Patients who did not develop DM during the observation period were censored (595 patients, 90%). Distant metastasis-free survival (DMFS) was calculated by the Kaplan-Meier method. Univariate comparisons between groups were performed using the log-rank test. A P value of 0.05 or less was considered statistically significant. Prognostic factors that were significant on univariate analysis were analyzed in Cox proportional hazards models for independent significance. Nonparametric qualitative and quantitative comparisons were performed using the Fisher exact test or Pearson χ² tests and the Mann-Whitney U test, respectively.

RESULTS

The median age of the patients was 60 years (range, 22-93). There were 488 male patients (74%) and 174 female patients (26%). The primary tumor was glottic in the majority of cases (366 patients, 55%), supraglottic in 265 patients (40%), transglottic in 30 patients (4%), and subglottic in 1 patient (<1%).

Sixty percent of patients presented with early T stage (T1/T2) primary tumors, and 169 (25%) had clinically palpable nodes in the neck (Table 1). The ipsilateral neck was involved in the majority of these patients (92%). Patients with supraglottic and transglottic tumors presented more commonly with clinically palpable nodes compared with those with glottic tumors (45% versus 26% versus 11%; P < 0.0001). Clinically palpable nodes were most commonly located at level II (56%) and level III (40%); more than 1 regional nodal level was involved in 68 patients (40%).

TABLE 1. Clinical T and N Staging of Patients at Initial Presentation

Single modality therapy was employed for the majority of early stage tumors (stage I and II; n = 337): radiation therapy (RT) in 217 (64%) and surgery in 104 (31%) patients. Patients with advanced stage disease (stage III and IV; n = 325) were mostly treated by a combined modality regimen: chemoradiation (n = 146 patients; 45%) or surgery followed by postoperative radiation (PORT) (n = 110; 34%); 21% were treated with a single modality (RT in 4% and surgery in 17%).

Figure 1 summarizes the management of the neck in the 662 patients. Of the 493 patients with clinically N0 disease, 38% received elective treatment to the neck, while the remaining 62% were observed under close clinical surveillance. The majority of patients whose necks were treated electively (139 of 188; 74%) had locally advanced primary tumors. Elective treatment to the neck consisted of RT in 21%, neck dissection alone in 18%, neck dissection plus PORT in 18%, and chemoradiation in 43%. The neck was appropriately treated in all 169 patients with palpable neck disease at presentation: neck dissection (10%), neck dissection plus PORT (46%), chemoradiation (40%), or RT alone (4%). One patient with a supraglottic N1 tumor received definitive RT before neck dissection, whereas 1 with a supraglottic N3 tumor was treated by neck dissection followed by postoperative chemoradiation.

FIGURE 1. Summary of the management of the neck. (CTRT, chemoradiation; RT, radiation therapy; PORT, postoperative radiation therapy; PND, post-treatment neck dissection).

The 5-year overall survival (OS), disease-specific survival (DSS), and recurrence-free survival (RFS) rates for the entire group were 67%, 83%, and 65%, respectively. Patients with clinically N0 necks had significantly better RFS, OS, and DSS compared with those who presented with palpable neck nodes (69% versus 55%, 73% versus 48%, and 88% versus 69%, respectively, P < 0.0001). Patients who presented with cN1 disease had significantly better RFS, OS, and DSS compared with those who had cN2 and cN3 disease at presentation (71% versus 41% versus 32%, 63% versus 37% versus 23%, and 86% versus 60% versus 25%, respectively). These differences were significant on both univariate and multivariate analysis (data not shown).

The median time to DM from index treatment was 18 months (range, 1-109). Distant metastases were recorded in 67 patients (10%). The lung was the most common site (69%), followed by bone (24%), liver (15%), skin (13%), mediastinal or axillary nodes (10%), and brain (3%). Nineteen patients (28%) failed at multiple distant sites, either simultaneously (37%) or sequentially (63%). The majority of the patients (46 of 67; 69%) had developed local and/or regional failure before developing distant metastases. The median interval from the first recurrence to DM in this group was 9 months (range, 1-107) and the lungs were the most common site of distant recurrence, which was comparable to patients who failed only at distant sites without locoregional recurrence (63% versus 76%; P = 0.4). Patients who developed locoregional failure had a significantly worse DMFS, OS, and DSS compared with patients who did not (73% versus 94%, 46% versus 75%, and 57% versus 95%, respectively; P < 0.0001). No significant difference in OS or DSS was detected between patients who developed DM alone compared with those that recurred locoregionally in addition (10% versus 17%; 9% versus 19%; P = 0.9). The 5-year OS and DSS in patients who developed DM were significantly worse compared with those who did not (15% versus 73% and 17% versus 93%, respectively; P < 0.0001). The median time to death from diagnosis of DM was 5 months (range, 1-160).

With a median follow-up of 60 months, the 5-year DMFS calculated by the Kaplan-Meier method was 88%. Even after accounting for the type of index treatment, the only significant predictor of worse DMFS on multivariate analysis was a higher clinical N-stage (P < 0.0001). Compared with the clinically N0 neck, the relative risk for DM was calculated to be 0.5 (95% CI, 0.2-1.4; P = NS) for cN1, 3.2 (95% CI, 1.7-5.9; P < 0.0001) for cN2, and 7.5 (95% CI, 3.1-17.9; P < 0.0001) for cN3 disease (Table 2). Elective treatment of the clinically N0 neck did not significantly alter DMFS compared with patients who underwent surveillance for cN0 necks (89% versus 93%; P = 0.1). Patients with clinically palpable disease at a single nodal level had a significantly better 5-year DMFS compared with those who had involvement of multiple levels (92% versus 58%; P < 0.0001). However, the location of the palpable nodes did not significantly alter DMFS (79% for upper neck versus 76% for lower neck; P = NS).

TABLE 2. Clinical Predictors of Distant Metastasis-Free Survival (DMFS) in All Patients Treated for SCCL

A subset analysis of DMFS examining the impact of chemotherapy according to the clinical nodal stage is presented in Table 3. Although the relatively small sample size in each of the subsets precludes any definite conclusions, it is noteworthy that patients who presented with clinically N2 necks had a better DMFS if they had received chemotherapy, compared with patients that were not treated with chemotherapy (83% versus 47%; P = 0.01). Similarly, patients with high volume neck disease (cN2 or cN3) had a better DMFS if they had received chemoradiation, compared with those who were treated with neck dissection ± PORT or RT alone (77% versus 50% versus 33%; P = 0.08). Patients treated with chemotherapy had a worse 5-year DSS compared with those that did not receive chemotherapy (76% versus 86%; P = 0.003). However, in the subset with cN2 disease, the trend was reversed and patients treated with chemotherapy had better DSS compared with those who were not (74% versus 38%; P = 0.1).

TABLE 3. Impact of Chemotherapy on Distant Metastasis-Free Survival (DMFS) Analyzed According to the Clinical N Stage at Presentation

The role of histopathologic covariates as predictors of DMFS was analyzed in a subset of patients who underwent primary surgery as index treatment (n = 268). As with the main analysis, Table 4 shows that the clinical N stage at presentation was the only independent predictor of DMFS in this subgroup of patients.

TABLE 4. Factors Affecting Distant Metastasis-Free Survival (DMFS) in Patients Treated With Primary Surgery (n = 268)

DISCUSSION

The rate of DM in our patients is comparable to that reported in recent literature.^4,5 Despite the relatively infrequent occurrence of DM in patients treated for SCCL, this population provides a unique opportunity for studying the impact of systemic therapy on DMFS because of the increasing use of organ-preserving chemoradiation approaches. More importantly, a systematic analysis of patients with SCCL is imperative for identification of patients who are at high risk of developing DM, so that this group may be selected for more aggressive therapy and then screened more thoroughly after completion of treatment. The problem with the existing literature is that most reports either include patients with tumors at other primary sites or fail to take into account the role of the type of initial therapy.^6,4 It is therefore difficult to extrapolate their observations in assessing risk of distant failure in an untreated patient diagnosed with SCCL. Some reports are based solely on patients who develop DM and attempt to identify risk factors in this selected population without considering the entire cohort of patients as a baseline denominator.^7,8,9 By analyzing all patients with SCCL treated at our institution during a relatively recent time period, we have sought to minimize these biases.

Our data demonstrate the direct relationship between the volume of clinically palpable regional nodal metastasis and the risk of developing DM in patients with SCCL. Patients with clinically N1 disease seem to be at similar risk as their N0 counterparts, whereas the risk increases theefold for N2 and more than sevenfold for N3 disease. The lack of significant difference in risk of DM between cN0 and cN1 disease may be attributable to several factors. Firstly, the difference in risk may indeed be slight so that the relatively small sample size in this retrospective analysis failed to detect it. On the other hand, 32% of the 54 patients who underwent therapeutic neck dissection for cN1 disease had pN0 or reactive lymph nodes, compared with 16% of the 32 cN2 and none of the 9 cN3 patients (P < 0.0001). Neck dissection was not performed as initial therapy in about a third of the 79 patients with cN1 disease, so that the actual pathologic status of these nodes could never be verified. It is conceivable that, as in the ones who underwent neck dissection, about a third of the patients who were treated nonsurgically also had pathologically negative nodes. Therefore, up to a third of cN1 patients may in fact have been overstaged on clinical examination. Finally, systemic chemotherapy was included in the primary treatment of 27% of patients presenting with cN1 disease, compared with 16% of those with cN0 necks (P < 0.0001). These observations may explain the lack of significantly increased risk of DM in patients who present with clinically low volume nodal disease. The impact of the volume of clinical nodal disease on prognosis has been examined in a cohort of patients with laryngeal and pharyngeal cancer treated with primary radiation therapy.⁶ The risk of DM was significantly higher for palpable neck nodes only above a critical, albeit arbitrary, volume of more than 100 cm³. Patients with palpable nodal volumes less than 100 cm³ could not be reliably stratified, and this observation mirrors our inability to demonstrate a difference in risk for DM between cN0 and cN1 patients.

As reported by other authors,^5,10 we have also shown that patients who present with clinically palpable metastases at multiple levels in the neck are at significantly higher risk of DM compared with those who have only a single nodal level involved. This finding is hardly surprising and is most likely a reflection of the aggressiveness of the primary tumor and/or the lead time to detection of metastatic disease. Other features of the nodal disease, such as location in the upper versus lower neck and presence of extracapsular nodal spread (ECS), have been reported as significant predictors of distant failure.^11,12 While the presence of ECS was a significant predictor on univariate analysis, we could not detect any significant difference in risk of distant failure relative to the location of nodal disease in the neck. It should however be noted that only 9% of the clinically involved nodes were located in the lower neck (levels IV and/or V) in our patients.

In the subset of our patients who underwent primary surgical treatment, we could not identify any pathologic feature that could predict worse DMFS independently of clinical nodal status. Poorly differentiated tumors had a worse DMFS compared with well-differentiated tumors on univariate analysis; however, like other variables examined, histologic grade of the primary tumor did not retain significance in multivariate analysis. Fortin et al¹³ reported that histologic grade of the primary tumor is a significant predictor of DM in a multi-institutional study of patients with SCCHN. None of their patients received systemic chemotherapy as part of their index treatment plan, and about 50% had tumors of anatomic locations other than the larynx. About a third of their patients were treated with primary surgery; after considering the 18% of patients who were excluded, it is unclear from their report whether they used the histologic grade of the biopsy or surgical specimen for analysis or indeed, what the level of concordance was between the two. Although the histologic grade of the tumor is almost always reportable on a biopsy specimen, this assessment is not only subject to observer variability, but it is also not always representative of the entire tumor. In any case, the utility of histopathologic features as predictors of outcome is limited by the fact that these are not available to the clinician for decision-making prior to initiating therapy, and also because they become irrelevant in patients who are treated with nonsurgical therapy such as CTRT.

Considering the elevated risk of DM in patients who present with high volume neck disease, it seems intuitive to consider adding aggressive systemic therapy to appropriate locoregional therapy in this group. It may however be argued that the development of distant metastases following successful locoregional treatment may represent disease that was always present as micrometastases and is now clinically apparent. There is some debate in the literature regarding the temporal relationship of locoregional recurrence to the development of distant metastases in patients with SCCHN. Authors who have studied patients who developed DM as the first site of failure put forth the argument that by excluding patients who fail locoregionally before developing distant failure, they are focusing on a cohort that likely harbored systemic micrometastases at initial diagnosis, but that subsequently became clinically apparent while the initial locoregional treatment kept the primary tumor at bay.^14,10 The problem with selecting this particular population for evaluating the impact of clinical nodal volume on DM is that it may introduce a bias by excluding patients with more aggressive disease that recurred locoregionally in spite of appropriate treatment. In including all patients that developed DM irrespective of their timing relative to locoregional failure and also using all patients treated for SCCL as a denominator, we have avoided making assumptions regarding the impact of systemic micrometastases on outcome because the relevance of this concept will not be apparent until a reliable means of identifying the presence of micrometastases becomes available.

If the issues outlined above are indeed valid, systemic chemotherapy should have the potential to reduce the development of DM. Obviously, our study was not designed to assess the role of systemic therapy in preventing DM in patients with SCCL. A review of the literature reveals conflicting data on the value of systemic chemotherapy (Table 5). The lack of convincing data in individual trials to support the efficacy of systemic chemotherapy in reducing DM may be due to the statistical inability to detect a marginal benefit in a small sample size. Most chemoradiation trials include patients who are heterogenous in terms of their characteristics, including tumor site and nodal stage. It is important to realize that factors such as the natural history, treatment methods, and prognostic predictors often depend on the site of the primary tumor and may complicate interpretation even of randomized trials that are otherwise properly conducted. Large numbers of patients would have to be assessed in site-specific trials to eliminate these biases. This is unlikely to happen for obvious reasons, and the next best available evidence comes from meta-analyses of randomized data. The MACH-NC group reported a meta-analysis of 602 patients randomized in 3 larynx preservation trials (VALCSG, GETTEC, and EORTC-24891) followed for a median of 5.7 years.¹⁵ Although there was significant heterogeneity between the 3 trials for the site of the primary tumor, they reported a significantly lower rate of DM as the first event in the chemotherapy group compared with the control arms (14% versus 19%; P = 0.001). The rate of DM after initial locoregional recurrence was, however, 3% in both arms.

TABLE 5. Summary of Randomized Data Evaluating the Impact of Systemic Chemotherapy on Distant Metastasis in Patients With Head and Neck Cancer

Our data demonstrate the impact of clinical nodal metastases at presentation on the development of not only distant but also locoregional failure. Additionally, patients with high volume clinical nodal disease who received systemic chemotherapy as part of their index management had a reduced rate of DM compared with those who did not, and a small subset with cN2 disease even displayed a trend toward better disease-specific survival. Our experience suggests that the benefit of systemic therapy in its current form may be evident in only a selected group of patients with cN2 disease. Patients who present with lesser disease have a considerably lower risk of DM so that the benefit of systemic therapy may be hard to detect in a study such as ours. On the other hand, the small group of patients with cN3 disease who were treated with surgery and/or PORT had better DMFS compared with those treated with chemotherapy. This may in part be due to the fact that patients with resectable neck disease were selected for surgery and PORT while the rest were treated with concurrent CTRT. Moreover, existing therapeutic regimens may be ineffective in either controlling systemic disease and/or prolonging survival for distant disease to become apparent in these patients with advanced tumors. Although these observations are based on only a small group of patients studied retrospectively, they highlight the need for assessing in a randomized setting the role of concurrent and/or maintenance systemic therapy added to locoregional modalities in selected high risk SCCL patients who present with high volume regional nodal disease (Fig. 2).

FIGURE 2. Some possible scenarios for integrating systemic chemotherapy in the management of squamous cell carcinoma of the larynx (SCCL).

CONCLUSION

Regardless of the index treatment modality, primary tumor site, or T stage, a higher clinical N stage at the time of presentation independently and significantly increases the risk of DM in patients with SCCL. Patients who present with clinically N2 disease are at threefold risk, and those with cN3 disease are at 8-fold risk of developing distant metastases during their course compared with clinically N0 patients.

Discussions

Dr. Marshall M. Urist (Birmingham, Alabama): We have already heard from a number of presenters at this meeting about the importance of lymph node status as a prognostic factor and even some debate about the therapeutic value of lymphadenectomy. Positive lymph nodes are an ominous sign, but what exactly does node status predict?

Dr. Shah and his colleagues from the Memorial Sloan-Kettering Cancer Center have taken on this formidable challenge by analyzing a large series of patients with carcinoma of the larynx. They have focused on the relationship between multiple pathologic and therapeutic factors compared with distant recurrence-free survival. They chose not to emphasize locoregional recurrence-free survival because their patients were treated with various nonrandomized protocols over this extended period of time.

The results show that the volume of node metastases was a significant independent predictor for the development of distant disease. Overall, this was not surprising. It was interesting, however, that the major increase in risk did not appear until patients developed N2 and N3 disease. This is distinctly different from other common malignancies where even a small nodal burden will result in a significant decrease in distant disease-free survival and overall survival.

With these findings the following questions are posed:

Was the lack of correlation with node positivity for N0 and N1 disease related to your lack of knowledge about the true nodal status for both patients with normal physical examinations of the neck and those with N1 disease?

There were a significant proportion of patients with N1 disease clinically who were found not to have metastases pathologically. Does this mean that there may be a role for sentinel lymph node biopsy for NO patients and the use of FNA to confirm the status for N1 patients?

How do you plan to use this prognostic tool in the management of your patients?

You have evidence that systemic therapy had a positive effect in reducing or delaying the appearance of distant disease. What new direction should we now take to confirm this finding and to extend it?

Dr. Jatin P. Shah (New York, New York): Indeed, ideally we would like to see patients who are at risk for developing metastases at N1 stage of disease, if we can identify those. Unfortunately the only available means of identifying early nodal involvement at present is an elective dissection of a select group of regional lymph nodes. Unlike the extremities and the trunk, the route of metastases in the head and neck are predictable. Progression of nodal metastases occurs in a sequential fashion and only a select group of lymph nodes are at risk for nodal metastases for each primary site. For example, for the larynx the sampling of lymph nodes at levels II and III and IV would be sufficient to pick up over 90% of lymph nodes in patients who are likely to metastasize or have micrometastases.

Another new imaging modality we are currently investigating is looking at the clinically negative neck with the PET scan (18-FDG Positron emission tomography). The problem with the PET scan is however that if the primary site and suspected nodal metastases are too close to each other, for example, in the hypopharynx, then the lymph nodes are difficult to identify and differentiate from the primary tumor.

Ultrasound examination with fine needle aspiration biopsy was promulgated in Europe by several investigators, and they have reported that it is a highly reliable diagnostic test to identify occult nodal metastases. If you have however a different ultrasonographer who does only 1 or 2 needle biopsies every other week, then the accuracy is not going to be that high. It is a highly technician dependent test and requires a dedicated sonographer who has interest in and is committed to the test. That is however difficult to find.

The information that identifies high risk patients for metastases based on the characteristics of the primary tumor and the extent of nodal metastases (N2-N3), is used to study prospective treatment programmers of systemic chemotherapy to reduce the risk of development of distant metastases by neoadjuvant chemotherapy is no longer a question. At least 3 prospective clinical trials published in the literature, including the very popular VA trial, indicated that the risk of distant metastasis is reduced by even a single course of neoadjuvant chemotherapy.

The problem is that we do not know how long to continue neoadjuvant chemotherapy. And many patients who complete their surgery and postoperative radiation therapy treatment program are simply exhausted and do not want to hear about any further treatment. In this group therefore compliance is poor. So we are going to have to come up with a treatment strategy which is easily tolerated by the patient and which can be given for a long period of time, eg, 6 months, although that is an arbitrary number, to see whether administration of adjuvant chemotherapy will lead to reduction in the development of distant metastases and eventually improvement in survival.

Dr. Philip E. Donahue (Chicago, Illinois): Dr. Shah, that was a very illuminating presentation. I wonder if I could ask you to focus on the N0 and N1 disease groups. Were there any observed differences in disease-free survival or in clinical course among those treated by either observation alone, chemotherapy, or radiation? Is there any overall risk to the wait and see approach, or should more aggressive treatment be done at the first chance?

Dr. Jatin P. Shah (New York, New York): That is a very good question. And that has been shown at various sites. Statistically we are unable to show improvement in survival by elective treatment of occult metastases to regional lymph nodes for any site.

However, we also know by clinical observation that metastatic disease does not necessarily progress in a sequential fashion from N0 to N1 to N2 to N3. If such was the case, then we could wait and treat all those who present with nodal metastases at N1. But many patients, even under our own surveillance, progress from their N0 status and come with multiple nodes within even a span of 1 month, and thus N2 disease.

We do have a study which brings that fact to light. We had 157 patients who underwent subsequent therapeutic neck dissection when they developed palpable neck metastases during observation following surgical treatment of their primary tumor. Sixty-six percent had a single palpable node (-3 cms), N, on clinical examination. However 68% had multiple metastatic nodes identified (N2b) in the surgical specimen. Thus the eventual prognosis worsens by the wait and watch policy.

So the current philosophy for squamous carcinomas of all head and neck sites is to electively treat the neck in those persons in whom the risk of micrometastasis exceeds about 1520%. And that is predicted by characteristics of the primary tumor and the location of the primary tumor.

The further back you go into the upper aerodigestive tract from the tips to the hypopharynx, the risk of nodal metastasis increases. So if you take into consideration on the primary site location, the T stage of the primary tumor, and the histology, as well as particularly the depth of invasion of the primary tumor, then you have pretty much of an idea as to who would be a candidate for elective treatment of the neck.

These patients would receive elective treatment of the neck by a limited selective dissection. In the case of oral cavity a supreomshyoid neck dissection. In the case of larynx and pharynx, selective jugular node dissection. If they are found to have multiple metastatic nodes or if there is extranodal spread of tumor, they will get postoperative radiation therapy. That particular approach has improved our local failure rate in the dissected neck, which was in the range of anywhere from 50 to 75%, down to 10% failure in the neck.