Tumor-Node-Metastasis Staging: The Early Years

In the original 1977 version of the AJCC manual, NMSCs were given one of four T classifications based on their extent of local spread (Table 2).⁵ Tumors limited to the epidermis and not exceeding 2 cm in diameter were called T1. Tumors wider than 2 cm but not exceeding 5 cm, or those with minimal invasion of the dermis, were called T2. Tumors wider than 5 cm, or those deeply invading the dermis, were called T3. Tumors involving cartilage, muscle, or bone were called T4. Next, all NMSCs were assigned an N classification based on their extent of regional nodal spread. Finally, each tumor was called M0 if there was no distant metastasis and otherwise labeled M1. In this first edition of the TNM system, no overall stage classification (I–IV) was assigned to NMSCs. Clinicians were advised to record the grade of cSCCs as well differentiated (G1), moderately differentiated (G2), poorly differentiated (G3), or very poorly differentiated (G4).⁵

Over the ensuing decades, the AJCC's TNM staging system was updated half a dozen times as the prognostication of NMSCs evolved. In their second edition in 1983, T staging remained the same, but the N categories were reconfigured: N1 indicated mobile ipsilateral nodes, N2 indicated mobile contralateral or bilateral nodes, and N3 indicated fixed nodes. At this time, the first-stage groupings were created: Stage I for localized disease, stage II for regional metastatic disease, and stage III if distant metastases were present.⁶

The third edition in 1988 brought further change: This was the first year that the AJCC's staging system was designed to be perfectly identical with the staging system of the Union Internationale Contre le Cancer, the multinational cooperative group based in Switzerland, so that there was now a single TNM system used worldwide.⁷ For the first time, SCCs of the eyelid, vulva, and penis were relegated to separate staging paradigms. For the remaining cutaneous malignancies, the T staging was modified so that extent of dermal invasion was no longer a consideration, but only the maximal tumor diameter distinguished T1, T2, and T3. The cutoffs between these groups remained 2 and 5 cm. N staging was now simplified to N0 vs N1, and the stage groupings were amended (Table 2).⁷ This version of the NMSC staging system remained the same in AJCC's fourth edition in 1992,⁸ in the fifth edition in 1997,⁹ and in the sixth edition in 2002.¹⁰

Tumor-node-metastasis Staging: The Present System

Prior to publishing the seventh edition of its Cancer Staging Manual in 2009, the AJCC convened an NMSC staging committee under the direction of Dr Arthur Sober. This committee was charged with revamping the staging of NMSCs so that, for the first time in history, it would be truly evidence-based. To this end, they eliminated the NMSC umbrella classification, which had previously encompassed 82 tumor types, and replaced it with a staging system specifically aimed at cSCC. Merkel cell carcinoma, on account of its uniquely aggressive behavior, was removed to its own staging system.

The present definition of a T1 cSCC is that it does not exceed 2 cm in greatest diameter and it possesses fewer than two high-risk features. High-risk features are defined as follows: Depth exceeding 2 mm; Clark level IV or V; perineural invasion; location on the ear or on the nonglabrous (i.e., hair-bearing) lip (recent literature is inconsistent when alluding to the inclusion of “nonglabrous lip” among the high-risk T factors in the seventh edition of AJCC staging. “Nonglabrous lip” is often mistranslated as “non-hair-bearing lip” or “vermilion lip.” The word glabrous derives from the Latin glaber, meaning smooth or hairless, and therefore, the nonglabrous lip is in fact the nonhairless lip, i.e., the hair-bearing lip, not the vermilion); and poorly differentiated or undifferentiated grade. A T2 cSCC either exceeds 2 cm in greatest diameter, or has more than one high-risk feature. A T3 cSCC is one that invades the maxilla, mandible, orbit, or temporal bone, and a T4 cSCC either invades the skeleton elsewhere or exhibits perineural invasion at the skull base.

Given that most cSCCs occur on the head and neck, the N classifications for cSCC were revised to coincide exactly with the N staging of head and neck mucosal cancer, and the overall stage groupings were revised (Table 2).

Even though the present staging rubric is more stringently evidence-based, it has garnered its share of criticisms.^3,11-14 The various subtypes of cSCC, from less aggressive histologies like verrucous carcinoma to more aggressive ones like spindle-cell carcinoma, are not distinguished. High-risk anatomic sites, such as the central face, temple, preauricular skin, and dorsal hands and feet are not mentioned.¹¹ Whether a patient is immunosuppressed, or whether her tumor arises in a scar, burn, sinus tract, or site of chronic inflammation, is not recorded.¹¹ Recurrent tumors are not considered separately. Many cSCCs never arrive at a final depth measurement, either because no residual tumor remains after initial biopsy, or because the chosen treatment modality is Mohs surgery, which focuses on surgical margins and often does not evaluate the excised tumor depth. Extranodal spread and micrometastases are not addressed. Finally, it seems illogical that appendicular bone invasion should automatically merit T4 status while facial bone invasion is only T3.¹¹

Alternative T Staging for Cutaneous Squamous Cell Carcinoma

In light of these concerns with AJCC staging, Jambusaria-Pahlajani et al¹⁵ set out to critically evaluate the performance of the current staging system on a retrospective cohort of 256 primary N0M0 cSCCs chosen to have at least one high-risk factor from among the following: Perineural or lymphovascular invasion, poorly differentiated histology, depth beyond subcutaneous fat, diameter at least 2 cm, location on the ear, or location on the lip. Eighty percent were treated with primary surgery, and 11% received postoperative radiation. During a median follow-up time of 44 months, the local recurrence rate was 8%, the nodal metastasis rate was 10%, and the disease-specific death rate was 5%. By the AJCC rubric, 54% of the cohorts were T1 and 44% were T2. As expected, given this distribution, 87% of incidences of recurrence, metastasis, or death occurred in the T2 group. The AJCC staging system did not perform well because so few tumors were T3 or T4, leaving a large T2 group which was widely heterogeneous in terms of outcomes.

Upon multivariate analysis of independent prognostic predictors of local recurrence, nodal metastasis, disease-specific mortality, and all-cause mortality, Jambusaria-Pahlajani et al¹⁵ proposed an alternative T staging system which would better stratify the T2 group. This system is based on the risk factors which they found to be strongly predictive of at least two of the above negative outcomes: Tumor diameter 2 cm or greater, poorly differentiated histology, perineural invasion, and invasion past subcutaneous fat. Stage T1 tumors have none of these four risk factors, T2a tumors have a single risk factor, T2b tumors have two to three risk factors, and T3 tumors have all four risk factors or bone invasion (AJCC T3/T4). By this alternative staging system, 52% of the cohorts fell into T1 and had local recurrence rates of 14%, nodal metastasis rates of 4%, and 0% disease-specific death; 26% of the cohorts were rebranded T2a and had local recurrence rates of 29%, nodal metastasis rates of 12%, and 0% disease-specific death; 19% of the cohorts were restaged T2b and had local recurrence rates of 43%, nodal metastasis rates of 72%, and 83% disease-specific death. Thus, the alternative staging system was able to reassign cSCCs to T stage groups with more distinct risks of negative outcomes when compared with AJCC staging (Table 3).

The Brigham and Women's Hospital (BWH) T staging system was subsequently validated on a series of 1,818 primary cSCCs, and again the AJCC T1/T2 outcomes were widely heterogeneous, while T3 and T4 lost their meaning as those groups were so small. Again the BWH stages were more distinctive, properly concentrating the majority of poor outcomes among T2b and T3.¹⁶

Alternative N Staging for Cutaneous Squamous Cell Carcinoma

As with T staging, the N staging of cSCC has stood to gain from a more rigorous grounding in evidence. Although fewer than 5% of cSCC metastasize, regional metastasis drops the 5-year disease-free survival rate to 70 to 75% even with combined modality treatment.¹⁷ When they metastasize, head and neck cSCC spread initially to the parotid and then to the neck.¹⁷ O'Brien et al¹⁸ hypothesized that parotid metastases and neck metastases would have independent prognostic significance. To test this, they prospectively collected 87 patients with cSCC involving the parotid gland and staged them according to the following system: P1 for a parotid node up to 3 cm, P2 for one parotid node between 3 and 6 cm or multiple parotid nodes, P3 for a parotid node larger than 6 cm or for involvement of the facial nerve or skull base, N1 for a neck node up to 3 cm, and N2 for any more extensive neck disease. On multivariate analysis, they found that P2 or P3 stage predicted for reduced control in the parotid, and that N2 predicted for decreased survival. O'Brien's system was then tested retrospectively on a group of 322 patients from six institutions, both Australian and North American. His new staging rubric was validated to some degree, as multivariate analysis showed that clinical P3 stage and pathologic N1/N2 stage independently influenced survival. Once again, however, there was no evidence that having three distinct P stages and three N stages was useful (Table 4).¹⁹

Building upon O'Brien's work, Forest et al²⁰ proposed a simplified N staging system for cSCC called N1S3, which counts the parotid as a basin of the neck. Patients with a single node 3 cm or smaller are N1S3 stage I; a single node greater than 3 cm or multiple nodes up to 3 cm qualify as N1S3 stage II; and patients with multiple nodes greater than 3 cm are N1S3 stage III. This system successfully stratifies those three groups of patients by locoregional control, by disease-specific survival, and by overall survival, all with a p-value <0.001. The authors were able to validate this on an external patient cohort as well. Their study also had implications for treatment: By showing that N1S3 stage I patients have a 5-year disease-specific survival of 90%, they could recommend that these patients be treated with parotidectomy and selective neck dissection alone and be spared the morbidity of adjuvant radiation, if there is no extracapsular spread.

The same group directly compared N1S3 N staging with AJCC N staging, using 603 patients from two prospective databases, and found the N1S3 system superior in terms of patient distribution, stratification, and parsimony. For instance, three of the AJCC groups (N2a, N2c, and N3) together contained less than 10% of all patients, whereas the smallest N1S3 group (III) contained 14%. The Kaplan–Meier disease-specific survival curves, which overlap each other in AJCC, separate widely in N1S3. The authors argued that accurate N staging is of particular importance in cSCC because nodal metastases often present late and it can be impossible to know which cutaneous primary led to nodal spread; furthermore, once a patient falls among the 5% of cSCC patients who develop nodal metastases, primary tumor factors no longer play as much of a role in the prognosis.²¹

Tumor-Node-Metastasis Staging: Next Year's Update

The eighth edition of the AJCC staging system will be implemented on January 1, 2018. One of the most prominent changes is that basal cell carcinomas, which had been included in the same AJCC staging system as cSCC from 1977 until 2009 until they were excluded starting in the seventh edition, have been brought under the same staging system as cSCC once more, along with all other NMSCs other than Merkel cell carcinoma. Conversely, all NMSCs occurring outside the head and neck have been excluded from the eighth-edition rubric and will fall under no AJCC staging system at all. The T staging criteria have been simplified, and the N staging criteria have been updated to reflect the poor prognosis conferred by extranodal extension (Table 5).