Analysis of the role of baseline PET/CT in a prospective study of unselected patients: The UK MCL biobank

Abstract Introduction: Mantle cell lymphoma (MCL) comprises 5-6% of NHL and has heterogenous clinical outcomes. There is growing recognition of an indolent subtype which current prognostic scores may not represent. 18FFDG PET/CT (PET) has prognostic value for many lymphoma subtypes. In particular, metabolic tumour volume (MTV) combining disease dissemination and proliferation, has emerged as an independent risk factor for progression free survival (PFS) and overall survival (OS) in other lymphoma types. PET can upstage MCL at diagnosis and is recommended in response assessment. There is conflicting evidence to support the prognostic role of radiomics in MCL, chiefly from retrospective series of aggressive disease. Method: The MCL Biobank Observational Study prospectively enrolled untreated MCL patients 16 years (y) old from 73 sites across the UK from 2014-19. Baseline characteristics, tissue samples, clinician assigned grouping: active monitoring (AM) or active treatment (AT), treatment and survival outcomes were collected. This study specifically investigates the prognostic role of PET in the MCL biobank cohort. PET were centrally reviewed using MIM (v7.3.4) semi-automated Lesion ID and source data. SUVmax, MTV and Total Lesion Glycolysis (TLG) were recorded by PERCIST criteria. Data was log-transformed. Medians of SUVmax, MTV and TLG were studied further. Lesions PERCIST were recorded as 0. Baseline demographics were compared using Pearson correlation (r), Chi-Square and Two-sided t-tests. Outcomes of time to first treatment (TTT), time to next treatment or death (TTNTD) and OS were assessed against radiomics by Kaplan Meier log-rank (KM), Cox regression, and Fine-Gray (FG). Competing risks were non-MCL deaths. Proportional hazards assumption was checked and met for all reported CR; log-log plots were assessed for FG. Results: 588 patients were recruited. Median follow-up was 5.2y. Baseline PET scans were obtained for 150/588. Median time from diagnosis to PET was 13 days (-110 to +67). All had PET avid disease, with wide variation in radiomics. Median PERCIST (liver/MBP)=3.9. Within the PET cohort, 45 underwent AM; 105 had AT (akin to the non-PET cohort). AM v AT differed in: B symptoms (29% v 47%, p=0.04), anaemia (38% v 59% p=0.02), high LDH (17% v 39% p=0.01), splenic involvement (29% v 66% p0.01) stage (stage I 13% v 4%, stage IV 56% v 78% p0.05) and Ki67 30% (89% v 53%, p=0.001). There was no difference in MIPI groupings. SUVmax distribution about the median (8.45) was equal (IQR 6.57). AM had a significantly lower median MTV (11.03, IQR 43.8) v AT (139.24, IQR 726.05). Correlations between radiomics and Ki67% or MIPI were weak. Within the PET cohort, MTV above the median was associated with immediate treatment (TTT 90d = 88% v 58%, p0.001) and successive treatment lines (23% v 11%, p0.05). 49% of deaths were non-MCL related. 139/150 received treatment: within AM = 34/45, median TTT 701d; within AT = 105/105, median TTT of 33d. Higher MTV predicts inferior outcomes. Median TTT for high MTV was 34d v low 85d, respectively, (95% confidence interval (95%CI) 27-41d v 0-258d, p0.01), FG Sub-distribution hazard ratio (SHR) 2.4, (95%CI 1.6-3.4, p0.001). Median TTNTD was 5y v Not Reached, p=0.047, SHR 2.3, (1.1-5, p=0.04). 5y OS was 48% v 60%, p=0.07, disease specific survival (DSS) SHR 3.1 (1.4-7, p0.01). TLG results were similar. Within the AM subgroup, higher MTV and TLG trended towards significance for shorter TTT: median TTT for high MTV was 511d v low 805d, p=0.5. Only 7/34 had 2nd line treatment. Radiomics did not correlate with TTNTD, DSS or OS. Differences in SUVmax high/low were not significantly associated with disease outcomes. Discussion: This study of unselected, untreated patients confirms the prognostic role of baseline PET in MCL. Higher MTV is predictive of shorter TTT and inferior TTNTD and OS, and a need for successive treatment lines. Patients assigned to AM were characterised by lower MTV which could potentially aid clinicians to identify patients appropriate for AM at diagnosis. This warrants further study, especially within the more indolent subset where limited patient numbers and MCL events likely precluded significant results. High SUVmax has been postulated to correlate with aggressive disease. We found no significant correlation between SUVmax and disease outcomes. Censoring competing risks had a significant impact on “p” values reflecting the specificity of PET for MCL, and a highly comorbid group.