The post-transplant lymphoproliferative disorders metagenomic shotgun microbial sequencing (PTLD-MSMS) study: clinical insights from the full cohort

Introduction: While PTLDs are rare, they still carry a high morbidity and mortality. Epstein-Barr virus (EBV) is causally associated with 50-80% of cases, but does not predict well either the presentation severity or the prognosis. The EBV-negative cases have no known etiology. This study was designed to collect a large quantity of tissues of PTLD to sequence viral genomes, especially EBV and anelloviruses, to determine how the sequence variation or presence might worsen the presentation or prognosis.
Methods: Over the period of this grant, the study centers (3 from USA, one from Australia) collected PTLD tissue samples (FFPE, frozen tissue, plasma). We also collected extensive clinical data, comprising 353 clinical only variables and 634 clinical pathology variables. Some samples were retrospective and retrieved from archives, and some were prospectively collected under informed consent. We analyzed the clinical data for multivariable associations and hazard ratios (HR) to 5-year mortality or event-free survival (EFS; mortality, repeat PTLD event, or allograft loss).
Results: The study centers accumulated 462 PTLD tissue samples (FFPE, frozen tissue, plasma) from 416 unique transplant recipients who had developed PTLD. The tissue requirement led to a higher proportion of late PTLD cases (more than 1-year post-transplant; n = 245/337; 73%). WHO histology distribution was predominantly monomorphic in 224, Burkitt in 8, Hodgkin in 15, polymorphic in 45. We excluded the 26 non-destructive PTLD cases from further analyses due to zero mortality and only 1 non-death event. On univariable testing, monomorphic had worse 5-year mortality but not worse 5-year EFS (Figure).

Multivariable associations to 5-year mortality were assessed from 266 patients (123 deaths; 46%) and showed significantly higher mortality in patients above age 18 at transplant (HR 2.4, p < 0.001), type of transplant either lung or stem cell (HRs 1.75 and 3.52, p = 0.062 or < 0.001, respectively), PTLD location in central nervous system (HR 4.33, p = 0.012). In contrast, lower mortality was seen with PTLD location in lymph nodes only (HR = 0.042, p 0.007) or rituximab eligible and used (HR 0.52, p =0.006). When assessing 5-year EFS in 257 patients (139 events; 54%), the same covariates showed nearly identical associations, with the addition of worse EFS in heart or intestinal transplant recipients (HRs 1.53 and 3.86, p = 0.049 and 0.012, respectively). In contrast, WHO histology classification (with Burkitt or T cell PTLD counted separately), time to PTLD, tumor EBV status and recipient gender did not associate to worse mortality or EFS.
Conclusions: To our knowledge, this largest ever tissue collection of PTLD also allows for a rich clinical dataset that validates some of the known worse prognostic factors (type of transplant, CNS location), but refutes others (Burkitt or T cell PTLD, late PTLD). We show novel findings of lower mortality if PTLD was restricted to lymph nodes only. Multivariable analyses to include the molecular data on EBV and anellovirus are ongoing. We also aim to develop a new clinical+molecular prognostic score, to determine if prior reported LMP1 mutations are validated in our cohort, and to assess if lytic transcripts of EBV are expressed in our tumor tissues (shown in some CNS tumors previously, but against the dogma that PTLD tissues contain EBV in the latent phase only).