Patient characteristics
Overall, a total of 377 B-iNHL patients, including FL, MZL, CLL/ SLL and LPL/WM, were diagnosed and treated during the January 2012 and December 2019 at our center. Patients who were diagnosed with grade 3b FL (n=9), non-newly diagnosed (n=44), with other hematological cancers combined (n=3) and had not any CRP or albumin record (n=78) were excluded. Finally, 243 patients, including FL (n=77), MZL (n=46), CLL/ SLL (n=91) and LPL/ WM (n=29), were found to meet study criteria and were involved in the analysis. The median age at diagnosis was 59 (range 19-85) years, with 133 (54.7%) males, and 110 (45.3%) females. The median follow-up time was 23 (range,1-83) months. 31 (12.8%) patients died, 45 (18.5%) patients had disease recurrence or died due to disease and 51 (21.3%) patients suffered disease progression or died due to disease by their most recent follow-up visit. (Table 1).
Serum CRP level and CAR cutoff selection
An optimal pretreatment CRP level cutoff value was determined by ROC analysis (area under the curve: 0.747, 95% CI: 0.631–0.894, specificity: 0.739, sensitivity: 0.739, p<0.001) (Figure 1A). This cutoff was 5.84 mg/L, also used in other time CRP level cohorts. Similarly, the optimal cutoff value of pretreatment CAR was 0.117 (area under the curve: 0.753, 95% CI: 0.634–0.872, specificity: 0.687, sensitivity: 0.783, p<0.001) (Figure 1B).
Association of CRP level, CRP kinetics and CAR with clinicopathological features
Pretreatment elevated CRP level was significantly correlated with clinical stage, reduced Hb level, high LDH level, elevated β2-MG level, disease types, IPI and treatment plan (p<0.05). The correlation between post-treatment elevated CRP level and age, reduced Hb level and disease types were observed (p<0.05). CRP kinetics was significantly correlated with age, clinical stage, reduced Hb level, elevated β2-MG level, disease types, IPI and treatment plan (p<0.05). Likewise, the correlation between CAR and clinical stage, reduced Hb level, elevated β2-MG level, disease types, IPI and treatment plan were observed (p<0.05). (Table1).
Prognostic value of pretreatment CRP level
Within the cohort of 243 patients, 222 had pretreatment CRP level record. Poor five-year OS (94.3% vs. 56.7%, p<0.001, log-rank test), DFS (88.0% vs. 33.9%, p<0.001, log-rank test) and PFS (78.3% vs. 40.6%, p<0.001, log-rank test) could be determined among patients whose pretreatment CRP level were elevated (Figure 2A-C).
Univariate Cox proportional analysis identified that pretreatment elevated CRP level was prognosticator for patients’ worse OS (HR: 6.546, 95% CI: 2.543-16.853, p<0.001), DFS (HR: 4.704, 95% CI: 2.305-9.601, p<0.001) and PFS (HR: 3.177, 95% CI: 1.694-5.961, p<0.001). For the multivariate analyses, in line with the recommended ratio of ≥5–9 events per variable, four variables, including pretreatment CRP level, clinical stage, PLT and β2-MG, was performed [37]. Since these four variables showed larger HR in the univariate model. We identified pretreatment elevated CRP level (HR: 5.110, 95% CI: 1.904-13.717, p=0.001) and reduced PLT (HR: 3.994, 95% CI: 1.635-9.754, p=0.002) as independent predictors for OS. Simultaneously, we also identified pretreatment elevated CRP level (HR: 3.767, 95% CI: 1.777-7.984, p=0.001), and reduced PLT (HR: 2.320, 95% CI: 1.071-5.023, p=0.033) were independent prognostic predictors for DFS. As well as, increased pretreatment elevated CRP level (HR: 2.606, 95% CI: 1.338-5.076, p=0.005) was proved predicting independently patients’ poor PFS. (Table 2).
Prognostic value of CRP level during treatment
In our study, elevated CRP level during treatment was not associated with five-year OS (89.7% vs. 82.4%, p=0.924, log-rank test), DFS (81.7% vs. 56.6%, p=0.378, log-rank test) and PFS (69.4% vs. 55.5%, p=0.586, log-rank test) (Figure 2D-F). Thus, univariate and multivariate Cox proportional analyses were not carried out among these patients.
Prognostic value of post-treatment CRP level
To evaluate the prognostic significance of post-treatment CRP level, 64 of 243 patients who had post-treatment CRP level record were involved in the analysis. Negative prognostic value of post-treatment elevated CRP level for five-year OS (81.7% vs. 39.4%, p<0.001, log-rank test) (Figure 2G) and DFS (56.4% vs.35.7%, p=0.020, log-rank test) (Figure 2H) could also be found. However, elevated post-treatment CRP level was not associated with five-year PFS (49.8% vs. 38.8%, p=0.058, log-rank test) (Figure 2I). So we just carried out univariate and multivariate Cox regression analyses of the main prognostic factors for OS and DFS in post-treatment CRP level subgroup.
Univariate Cox proportional analysis identified post-treatment elevated CRP level were associated with patients’ worse OS (HR: 6.728, 95% CI: 2.015-22.465, p=0.002) and DFS (HR: 2.601, 95% CI: 1.126-6.008, p=0.025). As post-treatment CRP level, PLT and β2-MG showed the larger HR in the univariate model, they were involved in the multivariate analyses. We found post-treatment elevated CRP level (HR: 5.826, 95% CI: 1.659-20.458, p=0.006) and reduced PLT (HR: 8.532, 95% CI: 2.065-35.241, p=0.003) independently predicted patients’ poor OS. In addition, we identified post-treatment elevated CRP level (HR: 2.384, 95% CI: 1.027-5.534, p=0.043) was independent predictor for DFS. (Table 3).
Prognostic value of CRP kinetics
Among 243 patients, highly significant association between CRP kinetics and poor five-year OS (80.0% vs. 82.8% vs. 47.1%, p<0.001, log-rank test), DFS (80.2% vs. 62.9% vs. 35.3%, p<0.001, log-rank test) and PFS (72.4% vs. 61.9% vs. 41.9%, p<0.001, log-rank test) could be found (Figure 2J-L).
Univariate Cox proportional analysis identified continuously elevated CRP level as prognosticator for patients’ worse OS (HR: 6.661, 95% CI: 2.775-15.988, p<0.001). Ever-elevated CRP level (HR: 3.483, 95% CI: 1.615-7.511, p=0.001 for DFS and HR: 2.638, 95% CI: 1.346-5.173, p=0.005 for PFS) and continuously elevated CRP level (HR: 5.232, 95% CI: 2.356-11.622, p<0.001 for DFS and HR: 3.663, 95% CI: 1.790-7.496, p<0.001 for PFS) were significantly associated with patients’ poor DFS and PFS. Multivariate analyses, including CRP kinetics, clinical stage, PLT and β2-MG were performed. We identified continuously elevated CRP level (HR: 6.461, 95% CI: 2.620-15.930, p<0.001) and reduced PLT (HR: 4.025, 95% CI: 1.791-9.049, p=0.001) could independently predict patients’ poor OS. What’s more, ever-elevated CRP level (HR: 2.425, 95% CI: 1.105-5.322, p=0.027), continuously elevated CRP level (HR: 4.748, 95% CI: 2.114-10.660, p<0.001) and elevated β2-MG level (HR: 2.100, 95% CI: 1.036-4.255, p=0.039) were found as independent predictors for DFS. Ever-elevated CRP level (HR: 2.086, 95% CI: 1.040-4.188, p=0.039) and continuously elevated CRP level (HR: 3.296, 95% CI: 1.594-6.818, p=0.001) also were found as independent predictors for PFS. (Table 4).
Prognostic value of pretreatment CAR
Within the cohort of 243 patients, 221 patients had pretreatment CAR record. Poor five-year OS (93.1% vs. 61.9%, p<0.001, log-rank test), DFS (87.3% vs. 42.8%, p<0.001, log-rank test) and PFS (77.2% vs. 48.8%, p=0.002, log-rank test) could be determined among these patients whose CAR were high (Figure 2M-O).
Univariate Cox proportional analysis identified pretreatment CAR were associated with patients’ poor OS (HR: 5.426, 95% CI: 2.018-13.962, p<0.001), DFS (HR: 3.856, 95% CI: 1.888-7.874, p<0.001) and PFS (HR: 2.591, 95% CI: 1.381-4.858, p=0.003). For the multivariate analyses, four variables, including pretreatment CAR, clinical stage, PLT and β2-MG, were performed. We identified pretreatment CAR (HR: 3.768, 95% CI: 1.415-10.034, p=0.008) and reduced PLT (HR: 3.475, 95% CI: 1.439-8.394, p=0.006) independently predicted poor OS. Simultaneously, pretreatment CAR (HR: 2.824, 95% CI: 1.336-5.971, p=0.007 for DFS and HR: 1.991, 95% CI: 1.021-3.882, p=0.043 for PFS) also was independent predictors for DFS and PFS. (Table 5).
Prognostic significance of CRP level, CRP kinetics and pretreatment CAR in the different diseases subgroups
Highly significant association between CRP level, CRP kinetics, CAR and poor OS, DFS, PFS (p<0.05, log-rank test) could be determined in CLL/ SLL subgroup. In FL cohort, a negative prognostic value of pretreatment elevated CRP level, CRP kinetics and CAR for OS, DFS, PFS (p<0.05, log-rank test) could be identified. However, elevated post-treatment CRP level was not associated with OS and DFS in FL cohorts. Furthermore, elevated CRP level, CRP kinetics and CAR were not associated with OS, DFS and PFS in MZL and LPL/ WM subgroups. (Table 6).