Background: The G8 screening tool has been developed to identify older cancer patients requiring a geriatric assessment for tailoring therapy. Little is known about its prognostic value, particularly by tumour site. An optimised version has been recently developed, but no prognostic information is available. We compared the prognostic value of both instruments overall and by tumour site.
Methods: Data were from a prospective cohort of cancer patients ≥70 years old referred to 1 of 6 French geriatric oncology clinics between 2007 and 2014 (n = 1333). Endpoints were overall 1- and 3-year survival. Cox proportional-hazards models were built to assess the predictive value of abnormal G8 and modified-G8 scores, based on published cut-offs or by classes of increasing risk. Sensitivity analyses involved adjusting for age, gender, treatment, metastasis, and tumour site (digestive, breast, urinary tract, prostate, other solid cancers, and haematological malignancies) and stratifying by tumour site and metastatic status.
Results: Abnormal scores were independently associated with overall 1-year survival: adjusted hazard ratio [aHR] = 4.3[G8]/4.9[modified-G8] and 3-year survival: aHR = 2.9/2.6; all p <0.0001. Associations persisted after stratifying by metastatic status and in most cancer sites (exceptions: colorectal (G8) and upper digestive cancer (both tools) [1-year analysis]; digestive cancers (both tools) [3-year analysis]). For both tools, classes of increasing risk showed a graded relationship with mortality (p < 0.0001).
Conclusions: Our results identified both abnormal G8 and modified-G8 scores as strong and consistent predictors of overall survival, regardless of metastatic status or tumour site. These findings strengthen the clinical utility of these instruments in the geriatric oncology setting.
Keywords: Cancer; Mortality; Prognosis; Screening.
Copyright © 2017 Elsevier Ltd. All rights reserved.