Patient Characteristics
The baseline characteristics of 60 patients are described in Table 1. Patients in our study group showed the male predominance with a median age of 49 years (range, 18 to 73 years). UAT-NTCL was diagnosed in 32 patients (53.3%) and NUAT-NTCL in 28 patients (46.7%). The majority of patients (81.7%) had elevated LDH levels. The presence of B symptoms was observed in 39 patients (65%) and 46 patients (76.7%) had two or more extranodal sites. Hemophagocytic histiocytes of BM were observed in twelve patients (20%). IPI, KPI, and PINK scores also suggested that the majority of patients (78.3%, 88.3%, and 71.7%, respectively) were in the high-risk category. The NLR ranged from 0.1 to 31.9 with a mean value of 3.0 and the PLR ranged from 1.5 to 1126.0 with a median value of 223.4. The best cut-off points for NLR and PLR were chosen to yield the highest difference in OS. The most discriminatory cut-offs for NLR and PLR were 3.2 and 133, respectively (Figure 1). Based on these cutoff values used in our subsequent statistical analyses, we classified the patients into the high and low NLR/PLR groups.
Clinicopathological factors associated with pre-NLR and -PLR
Potential relationships between pre-NLR/PLR and other clinico-pathological factors were shown in Table 2. At baseline, thirty-four patients (56.7%) had a low pre-NLR of < 3.2 and twenty-six patients (43.3%) had a high pre-NLR of ≥ 3.2. Twenty eight patients (40%) had a low pre-PLR of < 133 and forty two patients (60%) had a high pre-PLR of ≥ 133. No significant correlations were detected between pre-NLR and sex, age, UAT/NUAT presentation, LDH level, number of extranodal sites, BM involvement, Ann Arbor stage, IPI, KPI and PINK. However, high pre-NLR was found to be in significant correlation with B symptoms (p = .005) and ECOG PS (p = .007). As for low pre-PLR, it was significantly correlated with LDH level (p = .024), BM involvement (p = .017), and hemophagocytosis (p = .031), but not with any other factors examined here. Specifically, high pre-NLR was more likely to exhibit B symptoms and was significantly correlated with a poor ECOG PS. On the contrary, low pre-PLR was positively correlated with the presence of BM involvement and elevated LDH level.
Treatment response according to pre-NLR and -PLR
All patients received first-line chemotherapy consisting of ifosfamide, methotrexate, etoposide, and prednisolone (IMEP) plus L-asparaginase or pegaspargase (n=30); IMEP (n=21), cyclophosphamide, doxorubicin, vincristine, and prednisolone (CHOP, n=6); bleomycin, vincristine, and prednisolone (BVP, n=2); cyclophosphamide, vincristine, doxorubicin, bleomycin, procarbazine, and prednisolone (COPBLAM-V, n=1). The data on evaluation of treatment response to the initial therapy were available in 56 patients (93.3%) (Table 3). Complete remission was observed in 25 patients (44.6 %), partial remission in 9 patients (16.1 %), stable disease in 19 patients (33.9 %), and progressive disease in the remaining 3 patients (5.4 %), resulting in the objective response rate (ORR) of 60.7%. However, the objective response rate was not significantly different between two patient groups classified by pre-NLR (63.6% vs 56.5%, p = .592) and pre-PLR (53.3% vs 63.4%, p = .494). This indicates that both pre-NLR and -PLR might not serve as a predictive factor for response to chemotherapy in patients with advanced NTCL.
Prediction of survival
The median follow-up time was 15 months (range, 0.9 - 148.9 months). Forty-two patients (70%) died as of September 2016. Kaplan-Meier estimates of survival for patients with low or high pre-NLR/PLR levels are shown in Figure 2 and 3, respectively. The Kaplan-Meier curves for the survival revealed that patients with high pre-NLR had significantly worse OS (median OS, 6.6 vs 33.4 months, p = .017) and PFS (median PFS, 4.2 vs 7.5 months, p = .031). In contrast, low pre-PLR showed the significantly poor OS (median OS, 22.3 vs 6.2 months, p = .011) and PFS (median PFS, 6.9 vs 3.0 months, p = .008).
According to a Cox univariate proportional hazards analysis, the clinical factors associated with reduced OS were high pre-NLR (HR = 2.06, 95% CI 1.12-3.78; p = .020), low pre-PLR (HR = 2.24, 95% CI 1.18-4.26; p = .014), and ECOG PS ≥ 2 (HR = 2.14, 95% CI 1.15-3.98; p = .016) (Table 4). Multivariate analysis identified high pre-NLR (HR = 3.59, 95% CI 1.77-7.26; p < .001), low pre-PLR (HR = 3.78, 95% CI 1.79-7.95; p < .001), and number of extranodal sites ≥ 2 (HR = 3.39, 95% CI 1.33-8.62; p = .011) as independent poor prognostic factors for OS.
Regarding the PFS, the univariate analysis revealed that the clinical factors significantly associated with reduced survival were high pre-NLR (HR = 1.82, 95% CI 1.05-3.17; p = .034), low pre-PLR (HR = 2.16, 95% CI 1.20-3.90; p = .010), ECOG PS ≥ 2 (HR = 2.13, 95% CI 1.21-3.76; p = .009), high PINK score (HR = 1.98, 95% CI 1.03-3.80; p = .041), and elevated LDH level (HR = 2.60, 95% CI 1.10-6.16; p = .029). However, ECOG PS ≥ 2 (HR = 2.40, 95% CI 1.29-4.45; p = .005) was an only independent factor for PFS on a multivariate analysis.