Early Death in Multiple Myeloma. Analysis of Patients in Real-World Practice

Background. During the last years survival of myeloma patients has improve due mainly to the introduction of new drugs. Despite this, there are patients who die early: before two years from diagnosis. There are few data published of this group of patients with early mortality (EM) and most of them analyzing patients included in clinical trials.

Methods. We reviewed myeloma patients from our center from 1990 to 2017 and identified patients with EM. We recorded many clinical and biological variables, both at diagnosis and in evolution. We compared data of patients with EM with the rest and we analyze which variables confer worse prognosis of this group in a multivariate analysis. We also create a predictive model at diagnosis of patients with EM through a discriminant analysis.

Results. We identified 186 patients with less than two years of survival from diagnosis, 87 women and 99 men. Mean age was 70,3 years old (range 42-99). The most common cause of death was infection (42,33%) followed by progession (39,26%). We could compare data of these patients with those without EM from 1990 to 2009 (n=255) and we found that patients with EM were older (p=0,039), with more comorbidities (p=0,004) and worse ECOG (p<0,001). They had a disease more agressive and more evolved with worse bone disease (p=0,019), lower hemoglobine (p<0,001) and albumin (p<0,001); higher calcium (p=0,007), creatinine (p<0,001), LDH (p<0,001), β2microglobuline (p<0,001), immunoparesis (p=0,007) and percentage of plasma cells in bone marrow (p<0,001) and consequently worse Durie-Salmon (p<0,001) and ISS (p<0,001) prognosis score. We also found a worse response to first line treatment (p<0,001).

In a multivariate analysis, variables that confer a worst prognosis (overall survival) of patients with EM myeloma were β2-microglobuline ≥5,5mg/l (HR: 2,251, p=0,004), >4 points in Cumullative Illness Rating Scale (CIRS) (HR: 1,760, p=0,036) and progressive disease to first line tratment (HR: 2,230, p=0,007).

We create a mathematic predictive model at diagnosis through a discriminant analysis using serum calcium, ISS score and CIRS score to identify patients with EM. With this method we could classify correctly 68,4% of our patients.

Conclusions. In our real-world series patients with EM myeloma die mainly because infection or progression. They have worse basal situation (age, comorbidity, ECOG) an a more agressive and evolved disease as wel as worse response to treatment. We present a model to try to identify these patients at diagnosis, which could involve changing the management like giving antibiotic profilaxis to try to improve survival.