The patient characteristics of our study cohort are provided in Table 1. The median age was 59 years (IQR, 52–68), male to female ratio was 3:1, and 90% were of Middle Eastern (ME) descent (9% had South Asian and 1% had other descent). Comparatively, the median age at CLL diagnosis was 71 years in the Danish population-based cohort and the median age at treatment was 67 years in the multi-center U.S. study defining CLL-CI [8, 9]. At diagnosis, 222 patients (74%) presented with Binet stage A disease whereas 39 (13%) and 39 (13%) had Binet stage B and C, respectively. Among the 165 patients (55%) with evaluable FISH and/or TP53 mutation testing within 6 months of diagnosis, 8 (5%) had a TP53 aberration. The majority of patients (n = 194; 65%) underwent testing for IGHV mutational status, with 56% of these having unmutated IGHV.

Assessment of the CLL-CI-quantified comorbidity burden at diagnosis showed that 93 patients (31%) were low risk, 137 (46%) intermediate risk, and 70 (23%) high risk. Regarding the CIRS components that comprise CLL-CI, 150 (50%) of our patients had an endocrine comorbidity of at least moderate grade, 87 (29%) had a moderate or severe upper GI condition, and 49 (16%) had at least mild vascular comorbidities. For endocrine-metabolic comorbidities qualifying as moderate or severe by CIRS, 41% of KCCC patients had diabetes treated with oral agents or insulin, 10% had dyslipidaemia requiring medication, and 11% had other endocrine-related problems (e.g., thyroid disorders). The percentages of patients in our Middle Eastern cohort with other CIRS body system comorbidities (of mild, moderate, or severe grade) at CLL diagnosis were: 44% with hypertension, 9% with non-renal genitourinary, and 8% with musculoskeletal. The median total CIRS score was 5 (IQR, 2–8; range, 0–20). This is likely an underestimate as while medical comorbidities which impact disease management are well documented in our medical records, certain conditions (e.g., poor vision or hearing) are not routinely recorded, making accurate quantification of total CIRS challenging in retrospective analyses.

Median post-diagnosis follow-up was 49 months (95% confidence interval [CI], 42–61), 48 patients died (16%), and 139 (46%) received treatment for CLL, with treatment initiated at a median of 11 months (IQR, 3–36) after diagnosis. 54% (163 patients) had an EFS event during follow-up and the median EFS was 39 months (95%CI, 30–57) (Fig. 1A). Stratifying by CLL-CI revealed a statistically significant difference in EFS (log-rank p < 0.001), with median EFS of 88 (95%CI, 53-NA), 30 (95%CI, 27–48), and 24 (95%CI, 14–39) months in the low, intermediate, and high risk groups, respectively (Fig. 1B).

Event-free and overall survival curves. Event-free survival Kaplan-Meier curves of the entire cohort (A) and by CLL-CI risk group (B). EFS curves for each CLL-CI group adjusted for disease stage and TP53 aberration (via creation of an artificial dataset with balanced CLL-CI distribution) among the subset of patients with known TP53 status within 6 months of diagnosis (C). Overall survival Kaplan-Meier curves for each CLL-CI group (D)

Evaluating the association between individual CLL-CI components and EFS in our Middle Eastern cohort revealed that neither presence of moderate/severe endocrine comorbidities (HR = 1.24, p = 0.164) nor vascular conditions (HR = 1.05, p = 0.828) were significant predictors of EFS (Table 2). However, moderate or severe comorbidity involving the upper GI system was strongly associated with a shorter EFS in both a univariable (n = 300; HR = 2.05, p < 0.001) and a multivariable Cox model with disease stage and TP53 status as covariates (n = 165; HR = 2.36 [95%CI, 1.51–3.68], p < 0.001; c-index = 0.73).

Multivariable Cox regression applied to the 165 patients with evaluable TP53 status revealed that CLL-CI remained a significant predictor of EFS (intermediate vs. low risk: HR = 2.18 [95%CI, 1.22–3.90], p = 0.009; high vs. low risk: HR = 2.32 [95% CI, 1.23–4.37], p = 0.009; c-index = 0.73) after adjusting for disease stage (Binet A vs. Binet B/C) and the presence of del(17p) or TP53 mutation (Fig. 1C). As TP53 disruption was rare (5% of tested patients) and unknown in 45% of patients, we also examined the independent effect of CLL-CI in the 300-patient full cohort when adjusting for disease stage, the only non-CIRS feature significantly related to EFS and recorded in all patients (Table 2). We also separately analysed the 194 patients with known IGHV status where we adjusted for disease stage and IGHV. In these multivariable models, higher CLL-CI risk was a significant predictor of shorter EFS in both the full cohort (intermediate vs. low risk: HR = 1.79 [95%CI, 1.19–2.69], p = 0.005; high vs. low risk: HR = 2.55 [95% CI, 1.64–3.96], p < 0.001; c-index = 0.73) and the IGHV-known cohort (intermediate vs. low risk: HR = 1.74 [95%CI, 0.98–3.08], p = 0.057; high vs. low risk: HR = 2.38 [95% CI, 1.28–4.42], p = 0.006; c-index = 0.79).

CLL-CI was significantly associated with OS (log-rank p = 0.024) in the absence of covariates. Although the OS curve for low risk CLL-CI patients (median not reached) was well-segregated, there was minimal separation between curves for intermediate risk (median of 168 months) and high risk (median of 143 months) patients (Fig. 1D). When adjusting for age ≥ 65 at diagnosis (which was a significant predictor of OS in the univariable model, see Table 2), disease stage, and TP53 aberration, intermediate/high risk CLL-CI patients had shorter OS at borderline significance levels (n = 165; intermediate vs. low risk: HR = 4.59 [95%CI, 1.00-20.96], p = 0.049; high vs. low risk: HR = 3.99 [95%CI, 0.85–18.68], p = 0.079; c-index = 0.70). However, CLL-CI did not correlate with OS (intermediate vs. low risk: HR = 1.47, p = 0.532; high vs. low risk: HR = 1.17, p = 0.819) when adjusting for IGHV status (along with age ≥ 65 and disease stage) in the 194 evaluable patients with 23 observed deaths.

Interestingly, in contrast to our EFS findings, both moderate/severe endocrine comorbidities (HR = 3.18, p = 0.001) and any vascular condition (HR = 1.96, p = 0.039) were significant predictors of OS in the univariable setting. The presence of a vascular comorbidity was also an independent predictor of OS upon adjusting for other prognostic factors (HR = 2.49 [95%CI, 1.05–5.90], p = 0.038; c-index = 0.73). Furthermore, although moderate/severe upper GI comorbidity was an independent significant predictor of shorter EFS, it does not have a negative impact on OS (HR = 0.71 [95%CI, 0.29–1.72], p = 0.446; c-index = 0.65) and is thus inferior to CLL-CI in terms of predicting clinical outcome from the time of diagnosis.