Hemoglobin‐platelet index as a prognostic factor in patients with peripheral T‐cell lymphoma

Abstract Peripheral T‐cell lymphoma (PTCL) is a heterogeneous group of aggressive lymphomas with a poor prognosis. The International Prognostic Index (IPI) and the Prognostic Index for PTCL‐unspecified (PIT) is used to predict the prognosis of PTCL. The hemoglobin‐platelet index (HPI), based on anemia and thrombocytopenia status, is associated with the prognosis of diffuse large B‐cell lymphoma. However, its significance in terms of predicting the prognosis of PTCL has not been fully investigated. We herein retrospectively analyzed 100 patients with newly diagnosed PTCL in our department. At a median follow‐up of 3.2 years, the median progression‐free survival (PFS) and overall survival (OS) was 0.72 (95% confidence interval [CI]: 0.56–1.2) years and 2.0 (95% CI: 1.5–4.7) years, respectively. Multivariate analysis revealed that elevated lactic dehydrogenase (LDH) and hypoalbuminemia were independent adverse variables for PFS. The HPI showed significant predictive value for both PFS and OS. As a new prognostic model comprising the HPI, LDH, and albumin, the LA‐HPI allowed the stratification of patients into four distinct risk subgroups: low risk (zero risk factors), low‐intermediate risk (one risk factors), high‐intermediate risk (two or three risk factors), or high risk (four risk factors). The PFS and OS differed significantly among the patients by the LA‐HPI score. The LA‐HPI demonstrated better predictive performance compared to the IPI, PIT, and HPI. Our data demonstrated the prognostic utility of the HPI in patients with PTCL. The LA‐HPI, incorporating four readily obtainable parameters, exhibited superior performance compared to traditional indices.


INTRODUCTION
Peripheral T-cell lymphoma (PTCL) comprises a heterogeneous group of lymphoid malignancies constituting approximately 10% of non-Hodgkin's lymphoma cases (NHLs) in adults [1]. The most common subtypes are the so-called nodal PTCLs, including PTCL-not otherwise specified (PTCL-NOS), angioimmunoblastic T-cell lymphoma (AITL), and systemic anaplastic large cell lymphoma (ALCL) [2]. Although anaplastic lymphoma kinase (ALK)-positive ALCL has a relatively favorable prognosis in younger patients, older patients or those with a high International Prognostic Index (IPI) score (≥2) have a prognosis similar to that of patients with ALK-negative ALCL [3]. These subtypes are usually treated with cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP) or other CHOP-like regimens [4]. Although a small proportion of patients with PTCL might respond to, and even be cured with, conventional chemotherapy, the outcome of patients with PTCL is generally dismal [5]. Consolidation at first remission with highdose chemotherapy and autologous stem cell transplantation (ASCT) is often performed in young and fit patients in the hope of achieving better disease control. Unfortunately, however, most cases of PTCL eventually relapse after ASCT, and the superiority of this strategy has not been established [4,6,7]. Brentuximab vedotin (BV), an antibodydrug conjugate targeting CD30 plus cyclophosphamide, doxorubicin, and prednisone (BV+CHP) was approved after a randomized phase 3 trial (ECHELON-2). This approach resulted in a 5-year overall survival (OS) and progression-free survival (PFS) of 70.1% and 51.4%, respectively [8]. Although this novel agent has contributed to better survival outcomes in patients with PTCL, the outcomes are still less than satisfactory, and the search for better treatment options continues urgently.
Many previous studies have described the prognostic factors in the therapeutic response and survival of patients with PTCL. The IPI and the Prognostic Index for PTCL-unspecified (PIT) are the main instruments for assessing the prognosis of patients with PTCL [9,10]. Further, the predictive potential of several inflammatory parameters, such as C-reactive protein (CRP), serum ferritin, and β2-microglobulin, has been investigated [11][12][13], and the Glasgow Prognostic Score (GPS), an inflammation-based prognostic score based on CRP and albumin values, can also predict the prognosis of PTCL [14]. The controlling nutritional status (CONUT) score, a simplified nutritional index based on serum albumin, total cholesterol, and total lymphocyte values, is also reportedly an independent prognostic factor in patients with PTCL [15].
Several clinical markers using hematological parameters, such as platelet count and hemoglobin (Hb) level, have also been recognized as important prognostic factors for PTCL [16][17][18]. The hemoglobin-platelet index (HPI) is calculated using the Hb value and platelet count, and is also reportedly an independent prognostic factor for diffuse large B-cell lymphoma (DLBCL) [19]. However, its prognostic value for patients with PTCL has not been specifically investigated.
In this study, we aimed to investigate the prognostic factors associated with PTCL. Furthermore, we explored the predictive value of the HPI and assessed the potential for refinement of this index.

Patients
All

Prognostic scores in PTCL
The IPI and PIT scores were calculated as reported previously [9,10].
The IPI scores were calculated using age, serum lactate dehydrogenase (LDH), PS, Ann Arbor stage, and extranodal involvement at diagnosis. The PIT scores were based on age, PS, LDH, and bone marrow involvement. The HPI was calculated by assigning one point to the presence of anemia (Hb < 13 g/dL for male patients and Hb < 12 g/dL for female patients) or thrombocytopenia (platelets <100 × 10 9 /L).

Statistical analysis
OS was defined as the time from diagnosis to death or the latest followup. PFS was defined as the time from diagnosis to relapse or death from any cause or the latest follow-up. OS and PFS were estimated using the Kaplan-Meier method and were compared using univariate analysis with the log-rank test. Risk factors statistically significant on univariate analysis were included in multivariate analysis. The differences in the characteristics of the two groups were assessed using Fisher's exact test. The discriminatory power and prognostic performance of each prognostic index were assessed using Harrell's C-index and area under the curve (AUC), which was derived from the receiver operating characteristic (ROC) curve. All

Patient characteristics
In total, 100 patients meeting the criteria described above were included. The median follow-up period in the surviving patients was 3.2 years (range: 0.54-11.6 years). Table 1

Outcomes in the entire study cohort
In the total cohort, 63 patients experienced a relapse or pro-

Conventional prognostic scores of PTCL
The IPI and PIT significantly stratified PFS and OS ( Figure 3).
However, the differentiation of survival curves was suboptimal in some areas; for example, the IPI failed to demonstrate a significant distinction in PFS between the low-risk and low-intermediate-risk

Significance of the HPI as a prognostic factor
Patients with thrombocytopenia had significantly worse PFS and OS than those without thrombocytopenia ( Figure 4A,B). Similarly, patients with anemia had significantly worse PFS and OS than those without anemia ( Figure 4C,D). According to the HPI, 35 (35%) patients were classified as low-risk, 57 (57%) patients as intermediate risk, and eight (8.0%) patients as high risk. The HPI showed a significant association with PFS and OS ( Figure 5A,B).  Figure 6A). Then, we compared the predictive performance of these indices. The Harrell's C-index of LA-HPI for PFS was better than that of IPI, PIT, and HPI (0.662, 0.636, 0.629, and 0.575, respectively). Additionally, the predictive values of these prognostic indices were evaluated by the ROC curves ( Figure 6B). According to the 2-year PFS, the LA-HPI exhibited a superior AUC compared to the IPI, PIT, and HPI (0.749, 0.699, 0.681, and 0.687, respectively). To track the changes in AUC among these models throughout the follow-up period, timedependent ROC curves were also constructed. These curves revealed a consistent superiority of LA-HPI over the IPI, PIT, and HPI throughout the entire observation period ( Figure 6C). The LA-HPI was able to distinguish the PFS and OS between the PTCL-NOS (n = 31) ( Figure 7A,B) and AITL (n = 43) groups ( Figure 7C,D).

DISCUSSION
The present study investigated the prognostic value of the HPI in patients with PTCL and found the HPI to be an independent predictor of PFS and OS. We constructed a prognostic index by combining the HPI, LDH, and albumin, previously identified independent The prognostic value of the HPI was originally investigated in DLBCL patients [19]. Thrombocytopenia was reportedly associated with a poor prognosis in patients with DLBCL and PTCL [16,18,[25][26][27].
The presence of anemia was also associated with a poor prognosis in follicular lymphoma and is used regularly in clinical practice as an Index (FLIPI) and FLIPI2 [28,29]. Anemia was also related to a poor prognosis in patients with DLBCL and PTCL [17,25,30]. Although the pathogenesis of thrombocytopenia and anemia in lymphoma is multifactorial and not completely understood, it is assumed to result from an autoimmune-mediated process and involves the bone marrow and inflammatory factors [19,31]. Based on these previous researches illustrating the prognostic significance of thrombocytopenia and anemia in patients with PTCL, we aimed to evaluate the prognostic utility of the HPI in this study.
In our cohort, the HPI was able to stratify the PFS and OS in patients with PTCL. Furthermore, we developed a more powerful prognostic index by combining HPI, LDH, and albumin. LDH is thought to reflect the tumor burden in patients with lymphoma, including PTCL, and is one of the variables in the IPI and PIT [32]. Hypoalbuminemia was also shown to be a significant risk factor for increased mortality in patients with PTCL [33,34]. The high prognostic values of these factors may help enhance the predictive power of the LA-HPI beyond that of the HPI.
AITL often manifests systematic symptoms associated with immune dysregulation and shows positive results on autoimmune tests [35]. In particular, autoimmune cytopenia can be seen as an initial presentation of AITL [36]. A previous study identified the presence of thrombocytopenia as an important, predictive factor for a poor prognosis in AITL [37]. In the present study, the LA-HPI index had a statistically significant prognostic value in the subgroup of patients with AITL and was useful in predicting their clinical outcomes.
PTCL is a heterogeneous group comprising aggressive forms of NHL and has a poor outcome following anthracycline-based chemotherapy, even when consolidation with ASCT is performed [4,6,7]. Several recent studies focused on gene expression profiling and microenvironmental immune cell signatures in PTCL [39][40][41]. Although these studies have shed light on PTCL biology, which has led to improvements in the therapeutic approach, there is presently no definite evidence that these results have any direct bearing on clinical practice. Considering that most of the methods examined in these reports are expensive and require special equipment, prognostic and predictive markers that are inexpensive and easily available in daily practice are still needed. Because the hemoglobin level, platelet count, LDH, and albumin level can be easily and reproducibly measured in routine practice, the LA-HPI deserves further investigation.
The present study has some limitations. First, it was a retrospective study enrolling a relatively small cohort and was conducted at a single center; thus, there are likely to be some biases, including selection bias. However, the latter was mitigated by including consecutive patients over a defined period. Second, the study population was heterogeneous with respect to various histological subtypes and treatment strategies, which might affect the robustness of our statistical interpretation. However, the LA-HPI demonstrated good discrimination of survival outcomes in the subgroups of both AITL and PTCL-NOS. Third, we were unable to explore the biological basis for the association between the HPI and LA-HPI and the prognosis of PTCL patient, as previously demonstrated in the study examining the correlation between anemia/thrombocytopenia and interleukin-6 production in DLBCL cells [19]. Finally, our study was conducted in a single cohort, potentially limiting the generalizability of our findings to other populations. We could not perform validation in independent cohorts; therefore, future studied are necessary to confirm the prognostic significance of the HPI and LA-HPI in patients with PTCL.
Despite these limitations, the data presented herein are the first to suggest that the HPI is an important prognostic factor in PTCL.
Considering that the HPI and LA-HPI consist of readily obtainable parameters, they are very useful indices for use in the standard clinical setting.

CONCLUSION
In conclusion, the HPI, based on anemia and thrombocytopenia values, was a significant prognostic factor of PTCL. Moreover, the LA-HPI demonstrated good ability to discriminate survival outcomes in patients with PTCL using only four parameters that are easily available in routine clinical practice.