Abstract
Personalized medicine, including targeted and immunotherapy, has substantially changed the landscape of non-small cell lung cancer (NSCLC). About 30% of NSCLC tumors have targetable mutations, like EGFR, and 28% of tumors will have a PD-L1 score or tumor mutation burden high enough to warrant immunotherapy; a molecular test prior to therapy assesses eligibility. However, the congruence between testing and treatment, and the effect of treatment without testing on outcomes, is understudied. We extracted a cohort of NSCLC patients from the Surveillance, Epidemiology and End Results (SEER)-Medicare-linked data. The primary outcome was survival. Demographic and clinical characteristics of those receiving vs. not receiving a molecular test were compared, and a Kaplan-Meier curve along with a multivariable Cox proportional hazards regression assessed the association of survival with the receipt of molecular diagnostic testing, adjusting for sex, race, age, income tract, metro/urban/rural region, histology, and stage. There were 911 NSCLC patients treated with personalized therapy, of which 513 (56.3%) had received prior molecular testing. Black patients were less likely than White patients to receive testing (36.4% vs. 59.9%; p < 0.001). Testing was significantly more frequent in patients with fewer comorbidities (p < 0.001), adenocarcinoma (p = 0.004), tumors in the lower lobe (p = 0.037), or diagnosed at a later stage (p = 0.026). Only 39.9% of patients receiving EGFR inhibitors were initially tested for an EGFR mutation. Among NSCLC patients treated with personalized therapy, untested patients were at increased mortality risk compared to tested patients (median survival 8.22 vs. 12.79 months, p = 0.024). After adjustment, treated patients not tested beforehand had a significantly higher mortality risk (HRadj: 1.20; 95% CI: 1.04-1.40). Substantial incongruence between receipt of molecular testing and personalized therapies exists according to nationwide claims data, with only about half of treated patients having had an appropriate molecular diagnostic workup. This could explain, in part, the higher mortality among treated patients who did not undergo testing. Before expanding personalized therapies, further addressing these issues through standardized, reflexive testing protocols and improved clinician education is critical to optimizing the integration of molecular diagnostics into treatment planning.