Prognostic Role of the Red Blood Cell Distribution Width (RDW) in Hodgkin Lymphoma

Simple Summary The red blood cell distribution width (RDW) increases in inflammatory conditions and is described as having a prognostic role in different types of cancer. As Hodgkin lymphoma (HL) has a proinflammatory background, we aim to study the prognostic role of RDW in HL. We report in a large retrospective series of homogenously treated HL, for the first time, that RDW is a simple, cheap, and easily available prognostic factor in HL, that identifies a group with worse EFS, OS, and a higher potential incidence of secondary malignancies. RDW seems to be related to most adverse prognostic factors in HL and this may make RDW a good candidate to be included in current or new prognostic scores for HL. Abstract The red blood cell distribution width (RDW) is a parameter available from an automated blood count, which measures the degree of heterogeneity of erythrocyte volume and increases in inflammatory conditions. The prognostic role of RDW has been described in different types of cancers. Hodgkin lymphoma (HL) is a hematological malignancy, known to have a proinflammatory background. We aim to study the prognostic role of RDW in HL. We retrospectively analyzed 264 patients with HL from two hospitals in the Balearic Islands between 1990 and 2018. Higher levels of RDW were independently related to anemia, B-symptoms, and low albumin. In age ≥45 years, the presence of lymphopenia and higher RDW were independently associated with worse event-free survival (EFS) and overall survival (OS). Long-term incidence of secondary malignancies was significantly higher in patients with higher RDW, particularly lung cancer. In conclusion, we report for the first time that RDW is a simple, cheap, and easily available prognostic factor in HL that identifies a group with worse EFS, OS, and a higher potential incidence of secondary malignancies. RDW seems to be related to most adverse prognostic factors in HL, making RDW an excellent candidate to be included in prognostic scores for HL.


Introduction
Hodgkin lymphoma (HL) is a hematological malignancy characterized by few neoplastic cells called Reed-Sternberg inside an inflammatory microenvironment [1]. Standard therapy regimens cure approximately 80% of patients, but the other 20% will require salvage therapy [2]. Identifying factors that could improve the early detection of these refractory patients is very important to improve risk stratification and individualize treatment [3].
The red blood cell distribution width (RDW) is a simple blood test parameter that reflects the size diversity of red blood cells (anisocytosis)in peripheral blood and traditionally was used to study anemias [4]. In the last decade, higher levels of this parameter have been described as an adverse prognostic factorin cardiovascular diseases, inflammation, and cancer [5][6][7][8][9][10][11].
In cancer patients, higher values of RDW could be associated with a higher degree of inflammation. Increased levels of cytokines may modify iron metabolism by increasing levels of hepcidin and oxidative stress. Simultaneously, erythropoietin production is reduced, resultingin more anisocytosis and higher values of RDW [12,13].
However, there are scarce data for HL, where a proinflammatory background is a key eventin pathogenesis and physiopathology [25]. We aim to analyze the potential prognostic role of RDW in HL.

Patient Characteristics
A total number of 264 patients with classic HL homogeneously treated with ABVD +/− RT were retrospectively analyzed at the time of diagnosis, at Son Espases (n = 165) and Son Llatzer (n = 99) University Hospitalsin the Balearic Islands, between 1990 and 2018. The presenting features of the patients are shownin Table 1. The median age was 37 years (14-83 years), 52% of patients had an advanced stage, 16% had bulky disease, 28% had an Eastern Cooperative Oncology Group Performance Status (ECOG PS) >1, and 19% had an IPS >3. All prognostic factors of IPS are also shownin Table 1.

Analysis of the Prognostic Role of RDW
As shownin Table 1, the median RDWin the cohort was 13.9 (range, 10.6-23.9). To evaluate the ability of RDWin predicting a worse outcome, we considered the progression or death of any cause. As normal values changed with the center and the time, we standardized the values as a ratio of the upper normal valuein each center and time. Using a ROC analysis, we obtained an optimal cutoff of 0.95 of sRDW with an area under the curve of 0.64 (CI95%: 0.57-0.71) (p = 0.001) (Figure 1).

Response and Survival Analysis
With frontline therapy, 88% of the patients reached complete response (CR), 4% had a partial response (PR), and 8% a stable/progressive disease (SD/PD). With a median follow-up of 81 months (range, 11-352), six-year event-free survival (EFS) and overall survival (OS) were 74% (CI95%: 72-77) and 86% (CI95%: 84-88), respectively. Table 3 shows the univariate and multivariate survival analysisin which we included all variables found to be significantin the univariate analysis and sex as a potential confounding factor. Briefly, EFS was significantly influenced by age, AA stage, B-symptoms, ECOG PS, IPS, and all related prognostic factors, excluding sex. Among the alternative inflammatory biomarkers tested: RDW ( Figure 2), ESR, and RCP were also related to EFS. However, only age ≥ 45 years, sRDW > 0.95, and the presence of lymphopenia were independently associated with a worse EFS. CR tended to be higherin patients with sRDW ≤ 0.95: 92% vs. 84% (p = 0.051).     In the case of OS, univariate analysis showed the influence of age, B-symptoms, ECOG PS, IPS, and all related factors, excluding sex and leucocytes. Regarding alternative inflammatory biomarkers, only CRP and sRDW seemed to significantly influence OS. Multivariate analysis found the same three prognostic factors (age ≥ 45 years, sRDW > 0.95, and lymphopenia) independently associated with a worse OS, together with the male sex, that was includedin the analysis as a potential confounding variable (Table 3).
Furthermore, the long-term incidence of second malignancies was significantly higherin patients with sRDW > 0.95: 15 (12%) vs. 4 (3%) (p = 0.015). Particularly, the incidence of lung cancer was much higher (5% vs. 1%)in patients with sRDW > 0.95 (Table 4). Unfortunately, the low incidence of specific secondary malignancies does not allow us to draw more definitive conclusions regarding the specific role of sRDW or other factorsin the incidence of each type of malignancy. While anecdotal, all three patients with head and neck malignancies were previously treated with radiotherapy (2 with sRDW ≤ 0.95).
Using binary logistic regression, we studied the prognostic factors related to the incidence of secondary malignancies. In the univariate analysis we observed a significantly higher risk of secondary malignanciesin older patients (p = 0.013), with Hb < 10.5 g/dL (p = 0.013), and sRDW > 0.95 (p = 0.014), but not radiotherapy administration (p = 0.15). However, we performed a multivariate analysis including all significant prognostic factors from the univariate analysis as well as radiotherapy as a potential confounding factor, and we found that only older age (RR: 1.03; p = 0.018), sRDW > 0.95 (RR: 3.84; p = 0.047) and radiotherapy administration (RR: 3.81; p = 0.014) were independently related to a higher incidence of secondary malignancies.

Discussion
To our knowledge, this is the first published report about the main prognostic role of RDWin HL. We previously presented part of this data at the 58th Annual Meeting of the American Society of Hematology [25]. In our study, RDW strongly correlated with main prognostic factorsin HL, and sRDW > 0.95 is shown to be an independent adverse prognostic factor for EFS and OS.
RDW is an automatically measured index of the heterogeneity of the erythrocytes [4]. Traditionally, this parameter was used for the differential diagnosis of anemias [26]. In cancer, anemia could be present due to inflammation, and after treatment [27].
Physiological conditions that could increase the RDW levels include aging, erythropoietin, pregnancy, black ethnicity, and physical exercise [13]. In previous years, an increasein RDW levels was described as an adverse prognostic factor that increases mortalityin the general population, associated with many acute and chronic conditions,in which inflammation represents a critical factor, including metabolic, cardiovascular, and thrombotic disorders [13].
RDW,in cancer, reflects chronic inflammation and poor nutritional status [28]. Certain studies support that cytokines play a central rolein RDW, being associated with advanced stages and higher mortality. It has been related to different inflammatory markers such as interleukin-6, ESR, CRP, soluble tumor necrosis factor receptors I and II, and soluble transferrin receptor [12]. Elevated levels of proinflammatory cytokines led to inadequate production of erythropoietin, impaired erythrocyte maturation, a poor nutritional status (hypoalbuminemia), and increased levels of hepcidin and oxidative stress. These are different biological mechanisms that may lead to higher values of RDW [28].
RDW has been shown to confer a worse prognosisin many types of hematological malignancies, including lymphoproliferative and myeloproliferative disorders. In aggressive lymphomas, such as DLBCL, elevated levels of RDW at diagnosis were associated with worse ECOG PS, B-symptoms, and higher IPI. They predicted a poorer prognosis [15,22]. It has been described as an independent adverse prognostic factor at diagnosisin MCL, and RDW also improved the prognostic stratification based on the simplified Mantle Cell International Prognostic Index (sMIPI) [16]. In extranodal NK/T nasal-type lymphoma, high RDW at diagnosis has been associated with poorer clinical outcomesin patients treated with radiotherapy-based schedules [29].
In indolent lymphoproliferative disorders, such as multiple myeloma, RDW at diagnosis has been described to predict the outcome and correlate with response to therapy [17][18][19]21]. Furthermore,in CLL, elevated RDW levels at diagnosis were associated with adverse prognostic factors such as advanced disease [14]. In hairy cell leukemia, high RDW was associated with active disease both at diagnosis and after receiving treatment [30].
Regarding myeloid malignancies, a prognostic role at diagnosis has been describedin chronic myeloid leukemia as a biomarker for risk stratification and its ability to predict response to treatment [31]. In myelodysplastic syndromes, high RDW valuesin patients with less than 5% blasts at diagnosis, was an independent prognostic factor. Approximately 30% of patients classified as IPSS-R lower-risk showed similar outcomes to those with higher-risk IPSS-R [32,33].
Most variables independently influencing PFS and OS detect HL's immune relationships (sRDW > 0.95, lymphopenia and older age). RDW is more related to systemic inflammation, while lymphopenia may be more associated with immune dysfunction. Older age may be related to both situations.
In our series, we found a strong relationship between RDW and most prognostic factorsin HL. Multivariate analysis showed that a higher RDW was independently associated with anemia, higher CPR, and low albumin. However, the most important characteristic of RDW is that it is a cheap and easily available prognostic factor that may be obtained from automatic blood cell counts at the time of diagnosis.
More importantly, RDW was independently associated with EFS and OS. We also found a relationship with the long-term development of secondary malignancies, a critical adverse prognostic factorin a malignancy such as HL with good long-term survival. RDW is increasedin patients with a higher proinflammatory background, more prone to adverse events, and shorter PFS as well as with a higher incidence of cancer in general.
Some of the present study's limitations include a retrospective analysis performedin two different centers over almost 30 years. However, we tried to minimize these limitations by using a significant sample (n = 264) without selection bias, homogenously treated (ABVD), and by standardizing the RDW values.

Patients and Sample Selection
We selected patients with classic HL homogeneously treated with ABVD (adriamycin, bleomycin, vinblastine, and dacarbazine) +/− radiotherapy (RT) at Son Espases and Son Llatzer University Hospitalsin the Balearic Islands, between 1990 and 2018. To avoid selection bias, we selected patients from the databases of the Services of Pathology and Pharmacy. Those patients treated with different schemes were excluded. This study was approved by the Ethics Committee of the Balearic Islands with the number IB4071/19.

Clinical and Laboratory Prognostiz Factors
Clinical variables were obtained from medical records including main prognostic factorsin main prognostic indexesin HL: age, gender, Ann Arbor Stage, lactate dehydrogenase (LDH) and β-2 microglobulin (B2M) serum levels, extranodal sites, B-symptoms, Eastern Cooperative Oncology Group performance status (ECOG PS), bulky disease and main variables of automated blood counts. As RDW values have been obtained using different techniques and measurement systemsin different centers and periods, we standardized them using the normal reference values of each determination, generating a standardized RDW (sRDW).
Main prognostic scores, International Prognostic Score (IPS), and those from the European Organization for Research and Treatment of Cancer (EORTC) and the German Hodgkin Lymphoma Study Group (GHSG) were calculated. Response assessment was done using Cheson [34] or Lugano criteria [35]in the corresponding time periods.

Statistical Methods
Qualitative or binomial variables were expressed as frequencies and percentages. Comparisons between qualitative variables were made using the Fisher Exact Test or Chi-square. Comparisons between quantitative and qualitative variables were performed through non-parametric tests (U of Mann-Whitney or Kruskal-Wallis). Receiver operating curve (ROC) analysis was used to obtain and optimal sRDW cutoff for progression or death of any cause. The binary logistic regression was used to find out the risk factors associated with sRDW and those associated with a higher risk of secondary malignancies.
Time to event variables were estimated according to the Kaplan-Meier method, and the log-rank test performed comparisons between variables of interest. Multivariate analysis with the variables that were significantin the univariate analysis and potential confounding factors was carried out according to the Cox proportional hazard regression model. A backward stepwise Cox multivariate analysis was performed to determine factors independently associated with PFS and OS. All p values reported were 2-sided, and statistical significance was defined at p < 0.05. Statistical analysis was performed using a statistical package program (SPSS Inc, Chicago, IL, USA).

Conclusions
In conclusion, we report for the first time that RDW is a simple, cheap, and easily available prognostic factorin HL, that identifies a group with worse EFS, OS, and a higher potential incidence of secondary malignancies. RDW seems to be related to most adverse prognostic factorsin HL, making RDW an excellent candidate to be includedin prognostic scores for HL. We are currently exploring new prognostic scoresin HL including variables easy to obtain from the automated blood counts, such as RDW.