The authors queried SEER 18 (1973β2015) using ICD-O-3 codes 8980/3, 8981/3, and 9342/3 and identified 66 cases β the largest cohort ever assembled for this diagnosis. Incidence was 0.02 per million; the parotid gland dominated (52 cases, 78.8%); mean tumor size was 4.4 cm with 60% β₯4 cm; 17.2% presented with distant metastases. Survival was poor: 5-year OS 37.2%, 5-year DSS 62.2%, median survival 38 months (95% CI 21.9β54.1) .
These population-level survival figures are less catastrophic than historical series: the classic 12-case M. D. Anderson series reported 50% mortality within five years and mean survival of 3.6 years , and one report cited 0% 5-year survival β so the SEER estimates likely reflect a broad, unselected mix of de novo and carcinoma ex pleomorphic adenoma cases.
Strengths: First population-level study of this tumor; rigorous methods (Kaplan-Meier, Cox regression, SPSS 25.0); explicit ICD-O-3 inclusion criteria; honest limitations section acknowledging coding reliance and lack of pathologic re-review .
Weaknesses BGPT identifies: (1) All significant predictors came from univariate Cox models β the text claims a multivariate model was used, but Table III is univariate, so confounding by stage/size was never adjusted for; with only 66 patients and ~11 metastatic events, robust multivariable fitting is statistically implausible. (2) The total parotidectomy "survival advantage" (HR 0.129, CI 0.021β0.800) is almost certainly confounded: surgeons choose total parotidectomy for larger/advanced tumors, so finding an inverse association suggests selection bias or immortal time effects rather than causation. (3) Chemotherapy data conflate "none" with "unknown" (54/66), and the near-significant chemotherapy HR of 2.447 (P = .057) plausibly reflects confounding by indication β sickest patients received chemotherapy. (4) No recurrence data, no comorbidity/performance status, no cause-of-death granularity. (5) Rising absolute case counts may reflect better coding or population growth rather than true incidence change. The authors themselves flag most of these, which is to their credit .
Diagnostic distinction matters: sarcomatoid salivary duct carcinoma can mimic carcinosarcoma histologically , and immunohistochemistry may help discriminate lineages β p40 is strongly expressed in squamous/myoepithelial differentiation but absent in pulmonary carcinosarcoma , while E-cadherin loss correlates with malignant transformation in salivary tumors . The paper does not discuss any immunohistochemical validation strategy β a genuine gap given SEER coding-only diagnosis. What would change the conclusions: a prospective multi-institutional cohort with central pathology review showing no OS/DSS benefit of total parotidectomy, or significant PORT benefit, would falsify the key recommendations.
Know what changed, what holds up, and what remains uncertain. Every Friday. No ads.