Study Identifies Predictive Classifier for Intensive Treatment of LAHNC

October 26, 2020
Hannah Slater
Hannah Slater

According to this study, patients with locoregionally advanced head and neck cancer with a higher risk of cancer progression relative to competing mortality, defined by a higher ω score, selectively benefit from more intensive treatment.

Patients with locoregionally advanced head and neck cancer (LAHNC) with a higher risk of cancer progression relative to competing mortality, as reflected by a higher ω score, selectively benefit from more intensive treatment, according to a study published in Cancer.

Moving forward, researchers suggested that the inclusion of additional prognostic variables could further improve the selection of patients for optimal therapeutic intensity.

“We sought to develop and validate a model to predict the benefit of treatment intensification for patients with LAHNC by using a novel methodological approach called generalized competing event (GCE) regression and to compare this with a standard modeling approach for risk stratification,” the authors explained. “In contrast to models that predict patients’ overall survival (OS) or progression-free survival (PFS), the GCE method is designed to stratify patients according to their hazard for primary events (eg, cancer recurrence/ progression) versus competing events (eg, death from noncancer causes).”

In this study, researchers evaluated 22,339 patients with LAHNC who were treated in 81 randomized trials testing altered fractionation (Meta-Analysis of Radiotherapy in Squamous Cell Carcinomas of Head and Neck [MARCH] data set) or chemotherapy (Meta-Analysis of Chemotherapy in Head and Neck Cancer [MACH-NC] data set). The models used were initially trained with the patients in MARCH who were treated in a control arm consisting of 5480 patients, and patients were stratified by tertile according to the ω score, which quantified the relative hazard for cancer versus competing events. The classifier was then externally validated using the MACH-NC data set.

Of note, factors found to be associated with a higher ω score were a younger age, a better performance status, an oral cavity site, higher T and N categories, and a p16-negative/unknown status.

Overall, the effect of altered fractionation on overall survival (OS) was greater in patients with high ω scores (HR, 0.92; 95% CI, 0.85-0.99) and medium ω scores (HR, 0.91; 95% CI, 0.84-0.98) compared with those who had low ω scores (HR, 0.97; 95% CI, 0.90-1.05; P for interaction = .086). Similarly, the effect of chemotherapy on OS was significantly greater in patients with high ω scores (HR, 0.81; 95% CI, 0.75-0.88) and medium ω scores (HR, 0.86; 95% CI, 0.78-0.93) versus low ω scores (HR, 0.96; 95% CI, 0.86-1.08; P for interaction = .011).

“Note that the model that we developed to predict ω scores (independently of treatment) performed well across 2 separate populations and treatment types, and this is a more stringent test of model validity than had we developed separate models for each data set,” the authors wrote. “Moreover, models to stratify patients by [progression-free survival; PFS] did not show significantly different predicted treatment effects across strata. The ω score is simple to calculate and could influence clinical practice in situations where there is uncertainty about whether or not treatment intensification would be beneficial.”

Importantly though, this study was limited by low numbers of older patients and a lack of some known predictors, including p16 status, smoking history, and comorbidity; thus, the inclusion of these and other covariates would likely enhance the discriminatory power of ω score classifiers in the future. Additionally, radiation therapy techniques and supportive care changed markedly over the course of the study period, and this may have influenced the current results.

Reference:

Zakeri K, Rotolo F, Lacas B, et al. Predictive Classifier for Intensive Treatment of Head and Neck Cancer. Cancer. doi: 10.1002/cncr.33212