Progression-free survival at 24 months and subsequent survival of patients with extranodal NK/T-cell lymphoma: a China Lymphoma Collaborative Group (CLCG) study

Limited evidence supports the use of early endpoints to evaluate the success of initial treatment of extranodal NK/T-cell lymphoma (ENKTCL) in the modern era. We aim to analyze progression-free survival at 24 months (PFS24) and subsequent overall survival (OS) in a large-scale multicenter cohort of patients. 1790 patients were included from the China Lymphoma Collaborative Group (CLCG) database. Subsequent OS was defined from the time of PFS24 or progression within 24 months to death. OS was compared with age- and sex-matched general Chinese population using expected survival and standardized mortality ratio (SMR). Patients who did not achieve PFS24 had a median OS of 5.3 months after progression, with 5-year OS rate of 19.2% and the SMR of 71.4 (95% CI, 62.9–81.1). In contrast, 74% patients achieved PFS24, and the SMR after achieving PFS24 was 1.77 (95% CI, 1.34–2.34). The observed OS rate after PFS24 versus expected OS rate at 5 years was 92.2% versus 94.3%. Similarly, superior outcomes following PFS24 were observed in early-stage patients (5-year OS rate, 92.9%). Patients achieving PFS24 had excellent outcome, whereas patients exhibiting earlier progression had a poor survival. These marked differences suggest that PFS24 may be used for study design and risk stratification in ENKTCL.

Previous studies have demonstrated that event-free survival (EFS) at 24 months (EFS24) and progression-free survival (PFS) at 24 months (PFS24) are important milestones to stratify patients with diffuse large B-cell lymphoma (DLBCL) and peripheral T-cell lymphoma (PTCL); survival beyond these time points is similar or equivalent to that of the general population [26][27][28][29]. Normalization to the survival of a country-matched general population was recently proposed as an alternative to the use of "time to event" analysis for lymphoma. However, clinical characteristics and treatment strategies differ notably between lymphoma subtypes. The clinical significance of PFS24 in ENKTCL remains unknown, and the impact of achieving PFS24 on subsequent OS remains unaddressed. In the present study, we aimed to examine the timing of events, post-treatment milestones and subsequent OS in patients with ENKTCL, with reference to the general background population.

Eligibility criteria and study population
Patients diagnosed with ENKTCL between 2008 and 2016 were retrospectively reviewed using information from the China Lymphoma Collaborative Group (CLCG) database [5,9,13,21]. The eligibility criteria in this study included patients who had received non-ANT-based chemotherapy and/or radiotherapy. Patients who had received ANT-based or unknown regimen chemotherapy were excluded. A total of 1790 patients formed the study population. The institutional review boards ethically approved this project and waived the requirement for informed consent because of the de-identification of patient data.

Statistical methods
Progression-free survival (PFS) was defined as the time from the date of treatment to the first of either disease progression, relapse or death from any cause. Subsequent OS was defined as the time from achieving PFS24 (24 months after treatment) or time from progression in patients who did not achieve PFS24 (progression within 24 months of treatment) to death from any cause.
The estimated hazard rates provide the trajectory of progression and death overtime and were smoothed by the Epanechnikov kernel. PFS24 was defined as being alive and progression-free 24 months after initial therapy. PFS was also evaluated at other landmark time points (12 and 36 months) for sensitivity analysis. OS was compared with the age-and sex-matched general Chinese population via standardized mortality ratios (SMRs), and expected survival was predicted using a conditional approach via the 'survexp' function in R (package survival). Survival curves we analyzed using Kaplan-Meier method. A two-sided P value of < 0.05 was considered to indicate a statistically significant difference. Statistical analyses were performed using SPSS (version 22.0; IBM Inc.) and R (version 3.6.2; http://www.r-project.org/).

Patient characteristics and survival
The baseline clinical characteristics of the included patients are listed in Table 1

Annual hazard rate over time
After initial treatment, 82% of progression and 80% of mortalities occurred within 24 months. Consistently, smoothed hazard plots illustrated maximal annual death and progression hazards of 15.4% and 27.1%, respectively during the first year. These hazard rates decreased to 6.1% and 7.6%, respectively, at 2 years (Fig. 1a). From 3 years the annual hazards for progression and death were <5%. Thus, PFS at 24 months (PFS24) is a logistical cutoff time point for further evaluation.
Subsequently, outcomes were examined with regard to other clinical factors and risk groups. Despite a significant difference in OS of patients between treatment subgroups (Table 1), the OS rate of patients after achieving PFS24 were favorable compared with the background population ( Table 2). There was little difference in subsequent outcomes after progression regardless of clinical characteristics or risk-stratified grouping ( Table 2). PFS24 status was informative for subsequent OS in all patient subgroups and could be used as a prognostic indicator.

Sensitivity analysis of PFS24 vs. other time points
For sensitivity analysis, we examined outcomes according to other landmark time points of PFS, including 12 (PFS12), 24 (PFS24) and 36 (PFS36) months. Although the subsequent 5-year OS rate continued to increase from PFS12 (86.5%) to PFS24 (92.2%), there was little benefit in examining beyond PFS36 (93.0%) and no significant difference between the PFS-based OS and expected OS (Table 3). Furthermore, there was little difference in the 5-year OS rate after progression regardless of the time point chosen (17.1% at 12 months vs. 20.3% at 36 months). Similar results were obtained when considering early-stage and advanced-stage diseases.

Discussion
This is the first study to evaluate PFS-based endpoints among patients with ENKTCL, primarily treated with non-ANT-based chemotherapy and frontline radiotherapy. Using a large multicenter cohort of patient data collected from the CLCG database, we verified that PFS24 acts as a prognostic indicator of subsequent survival under current treatment strategies. Patients who were progression-free b Subsequent OS of patients who achieved PFS24 after initial treatment vs. expected OS based on age-, sex-, and country-matched general population data. c Subsequent OS of patients who did not achieve PFS24 after initial treatment vs. expected OS based on age-, sex-, and country-matched general population data.
within 24 months after initial treatment resulted in favorable long-term outcomes, with OS times not distinguishable from that of age-, sex-and country-matched populations. Conversely, patients who experienced disease progression within 24 months had poor prognoses. This finding suggests PFS24 as a dichotomous variable, providing a clear benchmark for evaluating the success of initial treatment and for designing clinical trials for ENKTCL. The introduction of asparaginase-or platinum-based chemotherapy regimens and early implementation of highdose E-ISRT for localized disease have been the most important advances in the treatment of ENKTCL [11][12][13][14][15][16][17][18][19][20][21]31]. Although phase I/II trials or retrospective studies have reported improved survival with the use of non-ANT-based chemotherapy, the optimal chemotherapy regimens remain to be identified by prospective trials. Early efficacy endpoints such as PFS and EFS are required to reduce the duration of evaluation of treatment regimens, to deliver optimal treatment strategies to the clinic sooner, and so that ineffective regimens can be abandoned without prolonged evaluation [32]. However, due to the rarity of ENKTCL and the heterogeneity of treatments used, limited patient numbers make formal surrogate endpoint analysis difficult. Normalization to the survival of a matched general population is important in rare diseases such as PTCL and ENKTCL [29]. In the present study, PFS24 is identified as a dichotomous variable that may allow individualized prognosis prediction for patients with ENKTCL, as well as inform patient counseling, clinical decision making and prospective clinical trial design.
In previous studies, we have demonstrated that improved locoregional control is associated with prolonged PFS and OS in early-stage ENKTCL [33], and that the survival probability increased and the hazards of failure decreased in a risk-dependent manner [34]. In the present study, we demonstrate that the estimated risk of disease progression and mortality was initially high, but decreased dramatically within the first 24 months after current standard treatment for ENKTCL (Fig. 1a). Given that the vast majority of ENKTCL-related events occur within the first 24 months, patients achieving PFS24 are highly likely to have a nearnormal life expectancy, with actual and expected 5-year OS rates of 92.2% and 94.3%. Sensitivity analysis, examining time points at 12, 24, and 36 months, indicated that there was little OS improvement at PFS36, whereas the PFS12 endpoint did not dichotomize the two groups as effectively a Subsequent OS of patients who achieved PFS24 after initial treatment vs. expected OS based on age-, sex-and country-matched general population data. b Subsequent OS of early-stage patients who did not achieve PFS24 vs. expected OS based on age-, sex-, and country-matched general population data. c Subsequent OS of advanced-stage patients who achieved PFS24 vs. expected OS based on age-, sex-, and country-matched general population data. d Subsequent OS of advanced-stage patients who did not achieve PFS24 vs. expected OS based on age-, sex-, and country-matched general population data.     as PFS24, regardless of stage. In consistence with the observations of the current study, other studies have reported PFS24 to be an important endpoint for DLBCL and PTCL, associated with subsequent long-term outcomes [28,29]. Regarding ENKTCL patients who failed to achieve PFS24, the median OS after progression was only 5.3 months, indicating that few patients who suffer early progression can be successfully treated. As observed by other studies [35,36], ENKTCL patients who suffer disease progression after initial treatment had extremely poor prognoses, regardless of stage or risk group. In a recent study of 179 patients with ENKTCL who suffered disease relapse or progression after initial non-ANT-based chemotherapy, the median OS was only 6.4 months [35]. Similarly, for patients who failed to achieve PFS24 or EFS24, the subsequent median OS was 7.2 months after ANT-based immunochemotherapy for DLBCL and 4.9 months after ANT-based chemotherapy for PTCL [28,29]. These findings indicate that PFS24 could be safely incorporated into routine clinical practice without detrimental effects on long-term disease control or survival outcomes in aggressive lymphomas under current standard treatments. PFS24 should be further validated in the context of randomized controlled trials as a potential early efficacy endpoint for patients with ENKTCL. Furthermore, there is an urgent requirement to identify patients at high risk of early progression who would benefit from innovative treatment strategies. Following the achievement of PFS24, research for ENKTCL survivors should focus on lessintensive surveillance, long-term treatment toxicity, quality of life, and other outcomes [34,37,38]. Previous studies, mainly using the randomized controlled trial (RCT) or real-world data, have demonstrated that the surrogate endpoints of EFS and PFS are strongly related to OS at both trial-and individual-level in different lymphoma subtypes [26-29, 32, 39-41]. Consistent with the present finding, the subsequent survival of patients with DLBCL or PTCL who achieved EFS24 or PFS24 is almost equal to that of the age-, sex-, and country-matched general population [26][27][28][29]32]. Furthermore, we have confirmed the association of PFS24 with OS in trial-and treatment armlevel in RCTs on DLBCL patients treated with immunochemotherapy; its association has been externally validated using the literature-based data from high-quality phase II and retrospective studies [40]. Similarly, in patients with follicular lymphoma and marginal zone lymphomas [39,41], early progression of disease within 24 months after initial treatment stratified subsequent OS and identifies a high-risk population. Consistent with the present study, these findings highlight the important role of PFS or EFS as early efficacy endpoint in designing prospective trials. The further work is needed to determine the prognostic factors for PFS24 in patients with ENKTCL.
The limitations of this study are as follows: first, the impact of salvage treatment on survival outcome after progression was not assessed, as it is beyond the scope of this study. Originally, our dataset was designed to collect only baseline characteristics, initial treatments, and events during follow-up. Second, these findings were mostly based on patients with early-stage disease, primarily treated with non-ANT-based regimens and radiotherapy. Extrapolation of these results when considering patients who underwent other treatments would be speculative. Third, despite the wide enrollment of ENKTCL patients primarily treated with non-ANT-based chemotherapy and upfront radiotherapy from a real-world CLCG database, more work is required to externally validate the correlation between PFS24 and OS using independent data from other large database, particularly from nonendemic areas. Furthermore, the retrospective design of this study does not allow for assessment of EFS as endpoint. The evaluation of EFS24 vs. PFS24 may be explored in a future study using CLCG data. The patients included in this study represent a young population with few comorbidities; longer follow-up and comparison to other collaboration data will give more insight into the importance of these endpoints.
In conclusion, assessment of PFS24 stratifies subsequent outcomes in patients with ENKTCL. Patients who achieve PFS24 have favorable subsequent survival regardless of stage and risk group. However, patients who suffer early progression have extremely poor prognoses. These results support the use of PFS24 as an efficacy endpoint, and should be considered in future studies evaluating novel therapies and biomarkers for risk stratification.
Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons license, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons license and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons. org/licenses/by/4.0/.