Longterm outcome of patients with clonal cytopenia of undetermined significance: 10 year follow-up of a large unselected cohort Free

Introduction

The long-term outcome of patients with clonal cytopenia of undetermined significance (CCUS) remains ill defined. Selected datasets show a very high prevalence of disease progression to a confirmed myeloid malignancy though these are predominantly retrospective datasets from tertiary referral centres. There is also increasing interest in the non-haematological effects of clonal haematopoiesis with regards to general health outcomes and there is a requirement to have long term outcome data on prospective unselected CCUS cohorts. We have previously reported 5-year follow-up data on such a cohort of 400 CCUS patients identified through a large haematopathology service. This study represents 10-year follow-up of this cohort including progression and survival dynamics over time, predictive models of both progression and survival and cause of death.

Methods

All patient samples (n=2083) referred for investigation of unexplained cytopenia or suspected MDS to the Haematological Malignancy Diagnostic Service (HMDS) in the UK over a 2-year period (July 2014-2016) were subjected to targeted myeloid panel sequencing of 27 genes. This was alongside the established diagnostic pathway of morphology, flow cytometry and cytogenetics. A total of 400 CCUS patients were identified during this analysis. Outcome data was captured on these patients including subsequent diagnoses, overall survival and cause of death (the latter data capture is ongoing). Both pathologists and clinicians were blinded to the results of this analysis removing diagnostic and treatment bias.

Results

Median follow-up of this cohort was 9.6 years. The cohort included 256 males and 144 females, and the majority had only 1 somatic mutation (205/400; 51%) with TET2, SRSF2 and DNMT3A mutations predominating (47%, 22%, 19% respectively). At the point of analysis 317 patients were deceased (79%) while 79 patients had progressed to a confirmed myeloid malignancy (20%). The majority of progressions were to MDS (36/79; 46%) or myelodysplastic/myeloproliferative (MDS/MPN) overlap syndromes (25/79; 32%). The cumulative hazard for overall survival was near linear meaning hazard was constant over time. In contrast, most progressions occurred within the first 5 years following diagnosis (66/79; 84%) with approximately piecewise constant hazard estimates reducing from 0.061 (CI 0.047, 0.076) to 0.026 (CI 0.012, 0.041) after 5.14 years. Importantly there was an absence of reported progressions during the COVID-19 pandemic lockdown period.

Confirming, that the majority of patients were dying as a consequence of CCUS, 5-year net survival was 63.3% (95%CI: 56.4-70.2) compared to 47.0% (95%CI: 42.0-51.8) 5-year overall survival. Time to death analysis confirmed that the number of mutated genes had a significant impact on survival when incorporated into age and sex adjusted models while number of mutations was not significant. A predictive model including age, sex (M), ASXL1, BCOR, TP53, maximum VAF (variant allele fraction) and >1 mutated gene had a concordance index of 0.67. Importantly maximum VAF had a protective effect on survival becoming significant only when number of mutated genes was also included, consistent with a dominant clone predicting a better outcome versus multiple lower-level clones.

Time to progression analysis, in contrast, showed an overwhelming effect of the number of mutated genes in predictive models. A model including number of mutated genes and sex had a concordance index of 0.79 with no improvement in predictive power when other variables, such as maximum VAF were included. Progression free survival analysis confirmed the strongest effects identified for both survival and progression, producing a model containing age, gender, maximum VAF and number of mutated genes with a concordance index of 0.65.

Conclusion

Long term outcome analysis of this unselected CCUS cohort has provided further clarity on the natural history of this condition. At 10 years the progression rate is much lower than previously reported, though this is likely impacted by the very high mortality rate. Progression events most commonly happen early and are driven by mutation number while in those with multiple mutated genes, maximum VAF had a protective effect on survival. Net survival estimates confirmed the majority of CCUS patients were dying as a consequence of their disease; specific causes will be further explored.