Prediction of Poor Outcome in Older Patients Receiving Chemotherapy for Malignant Lymphoma: A New Frailty Scoring

Introduction: Patients "clinically fit" to receive chemotherapy suffering from malignant hemopathies, are an heterogeneous population covering fit and vulnerable patients. Patients with geriatric syndromes and/or irreversible comorbidities are usually excluded from high dose chemotherapy. However, a reliable "frailty score" remains urgently needed to better define the vulnerable population that does not benefit from chemotherapy. In the literature, two clinical (functional decline and Mild Cognitive Impairment (MCI)) and two biological (anemia and inflammation) factors are frequently correlated with poor overall survival (OS) or chemotherapy-related toxicity.

Objective: To determine a clinico-biological tool for the screening of vulnerable patients with malignant hemopathies presenting unacceptable chemotherapy-related toxicity or disappointing result defined as a poor OS.

Methods: This prospective multicentric study was conducted in the institute 'Jules Bordet' (Brussels) and in the University Hospitals of Leuven (Leuven). A Comprehensive Geriatric Assessment (CGA) was performed to 251 consecutive patients (65-90yrs) with malignant hemopathies admitted to receive chemotherapy. Clinical data, biological parameters and causes of death were extracted from medical records. A screening tool composed of 0 to 4 of the prognostic factors (loss of functional autonomy (Activities of Daily Living scale [ADL]), MCI (Mini Mental State Examination [MMSE]), anemia [hemoglobin] and inflammation [CRP]) was applied to our population. Univariate and multivariate Cox proportional hazards model were used to predict OS.

Results: One hundred and eighty two patients were evaluable for all characteristics (NHL, n=105; CLL, n=20; MM, n=26; AML, n=17; ALL, n=6; LMMC, n=3, MDS, n=5). Eighty-three percent had a more favorable prognosis (NHL, CLL or MM) and fifty-five percent have a first diagnosis of cancer. A "frailty" scoring system (range 0-4) was developed, based on items we identified as predictive factors: functional decline (ADL<6, n=94), Mild Cognitive Impairment (MCI) (MMSE<27, n=51), anemia (HB<11g/dl, n=90) and inflammation (CRP≥2mg/l, n=149). The population was stratified into 3 groups: fit (score=0-1, n=56), vulnerable (score= 2, n=60) and "frail" (score= 3 or 4, n=66). The OS was 86% in fit, 60% in vulnerable (hazard ratio (HR)=3.29; 95% CI=1.48-7.33; P=.004) and 41% in "frail" patients (HR=5.87; 95% CI=2.74-12.59; P<.001). Causes of death remain disease-related in a majority of the patients (82%). In our largest group of older patients (NHL, n=105), the frailty scoring was also applied (ADL<6, n=48; MMSE<27, n=29; HB<11g/dl, n=36; CRP≥2mg/l, n=85): the OS was 87% in fit (n=45), 65% in vulnerable (n=31) (HR=2.94; 95% CI=1.11-7.96; P=.034) and 41% in "frail" patients (n=29) (HR=6.61; 95% CI=2.60-16.83; P<.001) and thus reliable in this specific population.

Conclusions: In our selected population of patients with malignant hemopathies and particularly in the group of NHL, "clinically fit" to receive chemotherapy, our "frailty score" helps clinician to predict a poor OS. This scoring detects unsuspected "frail" patients who may benefit from palliative care. Further prospective analyses in a larger population, are on going to refine the score in other malignant hemopathies in order to avoid overtreatment in these vulnerable older patients.