Leukocytosis and Thrombosis At Diagnosis Are Associated with Poor Survival in Polycythemia Vera: A Population-Based Study on 327 Patients

Abstract ²⁸¹⁹

Introduction Polycythemia Vera (PV) is a myeloproliferative neoplasm with a long life expectancy. Consequently, long-term follow-up studies that focus on survival of PV patients must be conducted in order to register a sufficient number of events to provide informative results and statistical analyses. Most of the published studies are single-institution that focused on patients referred to hospitals, which does not necessarily reflect the course of PV in the general population. In these studies, the median age at diagnosis is younger than usually noted for PV patients in population-based studies. Moreover, survival is generally based on observed mortality, a good but incomplete representation of overall mortality: given the old age (>65 years) of half of the patients at the time of diagnosis, more specific analyses such as relative survival (RS) should be performed, because it describes mortality directly associated with the disease, thus ruling out biases related to age or other illnesses.

Material and Methods All PV patients registered from 1980–2008 by two well-defined population registries, the Côte d'Or area, France and Gothenburg, Sweden were included. Relative survival (RS), defined as the ratio of the observed survival rates to the expected survival rate, was calculated. This was done using general mortality life tables based on age, sex and calendar year, and containing the probability of death (all causes of death combined) in people with the same characteristics as those in our cohort for each country. A multivariate analysis was performed using a proportional hazards regression model for excess mortality as defined by Esteve et al. Analyses were performed with STATA (v11) software, using strel command and Dickman's program for univariate and multivariate analyses.

Results A total of 327 PV patients (188 from Gothenburg and 139 from the Côte d'Or area; 175 females and 152 males) with a median age at diagnosis of 71 years (range: 21–95) were analysed. The median follow up was 11 years.

As a consequence of both older age at diagnosis and longer follow-up, the mortality rate was higher than in other reports, which underlines the differences between population-based and single-center studies. Nevertheless, this high rate of mortality allowed us to better analyse survival on a sufficiently large number of events (n=244) compared to previous studies. With a median survival of 6.6 years, 244 patients (137 females and 107 males) died at a median age of 81 years. The median survival was 17.5, 8.5 and 6.4 years for patients <50, 50–59 and ≥60 years old, respectively.

RS with 95% confidence interval was 93% [86; 96], 72% [62; 79] and 46% [33; 57] after 5, 10 and 20 years after diagnosis, respectively. RS reached 50% after 17 years. No difference in RS between Gothenburg and the Côte d'Or area was noted. Using univariate RS analysis, age >60 or >70 years (p =0.011 and p<0.001, respectively), hyperleukocytosis > WBC mean value (11 × 10⁹/L; p=0.04) or median value (13 × 10⁹/L; p=0.004) and a thrombosis at the time of diagnosis (p=0.07) were all found to be risk factors of poor survival. On the contrary, neither the haemoglobin level, the hematocrit nor the platelet count (as continuous or categorical variables) were significantly associated with worse survival. Finally, using multivariate analysis for RS, prognostic significance was maintained for age >60 or >70 years, hyperleukocytosis >11 or >13 × 10⁹/L and thrombosis at diagnosis. Our study provides data showing that, in RS, age >60 or >70 years is still a risk factor of survival: in other words, there is excess mortality due to PV per se in people >60 or >70years compared to younger patients.

Taking into account these 3 prognostic factors, a model for RS was constructed in order to define 3 groups of risk: high for patients with 2 or 3 risk factors (age >70 years, WBC >13 × 10⁹/L at diagnosis, thrombosis at diagnosis); medium if there was one risk factor, and low when there were no risk factors (Figure 1). Ten year relative survival was: 25% [9; 44], 59% [42; 72] and 78% [60; 89] in the high, intermediate and low-risk groups, respectively (p<0.001).

In conclusion, PV patients aged over 70 years and/or with hyperleukocytosis >13 × 10⁹/L and/or thrombosis at diagnosis have worse relative survival, underscoring the need to focus on the WBC as a therapeutic target, and the importance of paying particular attention to these patients.