The Brigham and Women's Hospital staging system, and why squamous cell carcinoma has two

**The problem it was built for.** In a cohort of 256 primary high-risk squamous cell carcinomas, outcomes for AJCC stages T2 to T4 were statistically indistinguishable, because fewer than 2 percent of the cohort reached T3 or T4, which then required bone invasion. The result was that 83 percent of nodal metastases and 92 percent of deaths happened in stage T2 (Jambusaria-Pahlajani 2013). A stage that contains almost everyone and almost every bad outcome cannot be used to decide who needs extra treatment.

**How it works.** The alternative counts four independent risk factors: poor differentiation, perineural invasion, tumour diameter of 2 cm or more, and invasion beyond the subcutaneous fat. No factors is T1, one factor is T2a, two or three factors is T2b, and four factors or bone invasion is T3. In the derivation cohort T2b tumours were only 19 percent of cases but accounted for 72 percent of nodal metastases and 83 percent of deaths from the disease.

**How it holds up.** In 1,818 primary tumours from 2000 to 2009, AJCC and UICC T3 and T4 were indistinct, with overlapping confidence intervals, and were very rare, 0.3 percent and 3 percent of the cohort respectively; most poor outcomes fell in the low stages, 86 percent for AJCC and 70 percent for UICC. Under the Boston system only 5 percent of tumours were high stage, T2b or T3, and they carried 60 percent of poor outcomes, 70 percent of nodal metastases and 83 percent of disease-specific deaths. Ten-year cumulative incidences in low-stage tumours were 1.4 percent local recurrence, 0.6 percent nodal metastasis and 0.2 percent death from the cancer; in high-stage tumours, 24 percent, 24 percent and 16 percent (Karia 2014). Against the eighth edition, which was a real improvement on the seventh, the comparison is closer but the same in shape: AJCC 8 T3 and T4 were 18 percent of 680 head and neck tumours and the Boston T2b and T3 were 9 percent, and both captured about 71 percent of metastases and 85 to 92 percent of deaths; the Boston system had higher specificity, 93 percent, and higher positive predictive value, 30 percent, and better C statistics for nodal metastasis and disease-specific death, with no difference for local recurrence or overall survival (Ruiz 2019). The authors' conclusion was that AJCC 8's failure to separate T2 from T3 leaves a 23 percent group at significant risk, too large for routine nodal staging or adjuvant treatment.

**Where Britain stands.** UK reports stage against UICC (see `tnm-skin-carcinoma`) and stratify risk separately, and the Royal College of Pathologists has taken an explicit and dissenting position on the risk bands built on the Boston system. It sets out its own numerical thresholds: an adverse consequence rate, counting recurrence, nodal spread, systemic spread or death, below 5 percent is low risk, 5 to 20 percent is high risk and over 20 percent is very high risk; and any squamous cell carcinoma under 20 mm across or under 2 mm thick with no additional factors is, in its words, of negligible rather than merely low risk. A proposal derived from the Boston system calls 5 to 20 percent intermediate and over 20 percent high. The RCPath judges that restricting the word high risk to over 20 percent is inappropriate, that those intermediate cases are better called high risk and the high-risk ones very high risk, and that the proposal 'could potentially leave some serious intermediate cases under-rated in terms of risk and thereby receive inappropriate clinical management' (G124). Its final position is that risk stratification is better done by the treating clinician or the skin cancer multidisciplinary team than recorded as a core item in a pathology report at all.