Prognostic Value of Neuro-Psychological and Biological Factors in Clinically Fit Older Patients with Hematological Malignancies

BACKGROUND: A Comprehensive Geriatric Assessment (CGA) is recommended to detect vulnerable cancer patients for whom chemotherapy may lead to severe impairment on functionality, quality of life, or survival. Although CGA is useful for a better management of older patients with unsuspected problems, little is known about the reliability of CGA - or G8 screening tool - to optimize the therapeutic approach in a specific patient with a malignant hemopathy. Particularly, the prognostic value of cognitive functions, psychological status and other selected biological factors in older patients with hematological malignancies admitted to receive chemotherapy, are poorly investigated.

AIM: To assess in clinically fit (go and slow go) patients with malignant hemopathies, the reliability of G8 screening and CGA as a prognostic tool to predict overall survival (OS) and particularly, the predictive value of neuro-psychological and biological factors, in terms of one-year OS.

POPULATION AND METHODS: Between Oct 2010 and June 2013, G8 and CGA were proposed to 107 consecutive older patients (65+-yrs) with hematological malignancies, admitted to the hospital and judged clinically fit, by their refereeing physicians, to receive chemotherapy. An initial full-dose or reduced-dose chemotherapy has been administrated to these patients according to a multidisciplinary team decision. The nutritional status was screened by MNA and the cognitive functions by MMS and MoCA

RESULTS: Ninety patients -completely assessed for G8 and CGA- were evaluable for one year OS; Non Hodgkin’s Lymphoma (53%), Multiple myeloma (16%), Acute Myeloid Leukemia and Myelodysplastic Syndrome (15%), Chronic Lymphocytic Leukemia (14%) and Myeloproliferative Neoplasms (2%). Median age is 74 (65-89) yrs. 56% are males. 72% and 80%, were considered as “vulnerable” when evaluated with G8 (≤14.5) or CGA (≥2 impairments), respectively. The area under ROC-curve of G8 compared to CGA was 0.749 ± 0.051 (sensibility = 79.2% and specificity =55.6%). During the first year of follow-up, 28% (n=22) of older patients treated for hematological malignancies died. The leading cause of death (82%) was indeed the disease progression. Retrospectively, neither the G8 nor the CGA total score had an impact with the initial treatment choice and were not predictive for one-year OS. In addition, regarding patients characteristics, in a multivariate analysis (Cox proportional hazard model), age (p=0.015; HR=1.12) and diagnosis (p<0.001; HR=7.2), were major prognostic factors in terms of OS. Regarding specific CGA items, a multivariate analysis concluded that unsuspected cognitive impairment (evaluated with a MMS and a MoCA) had the strongest prognostic value (p=0.03; HR=3.560) in terms of OS.

DISCUSSION: In our clinically fit hematological patients, a poor G8 or CGA score does not translate into a worse survival, more likely because of two reasons: the disease itself remains the major cause of death and malnutrition, a reversible impairment after effective treatments, strongly impacts on G8 and CGA. The objective of an effective screening tool should thus be the identification of the population who could benefit from optimal treatment combined with an comprehensive management of unsuspected specific problems.

CONCLUSIONS: Our observations in “clinically fit” older hematological patients suggest that 1) the major cause of death is the disease itself and the priority should be to identify all patients susceptible to benefit from full dose chemotherapy 2) a poor G8 screening or an abnormal CGA are not specific enough and does not translate into a worse one year OS 3) some specific factors in the CGA (such as neuro-cognitive impairments) are highly correlated with one-year OS 4) age remain a continuous variable with a poor prognostic factor. Prospective trials are needed to further determine whether the combination of optimal treatment and improved management of impaired functions, including cognition, could lead to a better OS.