Treatment Outcomes and Relative Dose Intensity of Chemotherapy in Slovene Patients With Advanced Hodgkin Lymphoma

This retrospective study was undertaken to investigate the association of relative dose intensity (RDI) with the outcome of Hodgkin lymphoma (HL) patients with advanced stage disease receiving ABVD (doxorubicin, bleomycin, vinblastine, dacarbazine) and escalated BEACOPP regimen (bleomycin, etoposide, doxorubicin, cyclophosphamide, vincristine, procarbazine, prednisone). A total of 114 HL patients treated between 2004 and 2013 were enrolled for evaluation. RDI calculations were based on a Hryniuk's model. The association of variables with overall survival (OS) and progression-free survival (PFS) was analysed using univariate and multivariate Cox proportional hazards models. The median age of patients was 39 years, majority of patients were males and had stage IV disease. Fifty-four patients received ABVD and 60 received BEACOPP chemotherapy with 24 and 4 deaths, respectively. Patients in BEACOPP group were signicantly younger with lower Charlson comorbidity index (CCI) in comparison with ABVD group, making the comparison of groups impossible. In ABVD group, RDI was not signicantly associated with OS (p=0.590) or PFS (p=0.354) in a multivariate model where age was controlled. The low number of events prevented the analysis in the BEACOPP group. Patients' age was strongly associated with both OS and PFS: all statistically signicant predictors for OS and PFS from univariate analyses (chemotherapy regimen, CCI, RDI) lost its effect in multivariate analyses where age was controlled. Based on our observations, we can conclude that RDI is not associated with the OS or PFS after the age is controlled, neither in all patients combined nor in individual chemotherapy groups.


Introduction
Hodgkin lymphoma (HL) is a unique lymphoid neoplasm characterized by malignant Reed-Sternberg cells in an in ammatory background [1]. It has a distinctive bimodal age distribution with peaks around the second and sixth decade of life, but the incidence varies with the histological subtype and geography [2,3]. Average annual observed number of new HL cases in Slovenia between 2004 and 2013 was 46.8 [4].
Staging is based on the Lugano classi cation, which is derived from the older Ann Arbour classi cation system [5]. Afterwards, patients are assigned to one of the three categories -limited, intermediate and advanced stages, based on which the treatment is selected [6]. Combination of chemo-and radiotherapy are the backbone of classical HL treatment, particularly in early stages, in late stages radiotherapy is reserved to consolidate partial remission. Early stages of HL, comprising limited and intermediate stages, are generally treated with ABVD chemotherapy regimen (doxorubicin, bleomycin, vinblastine, dacarbazine) with involved-site radiation therapy. Advanced stages (stage IIB, III and IV) are usually treated with chemotherapy, and radiation therapy is used exclusively as consolidation for selected patients with partial remission [7][8][9][10][11]. Chemotherapy regimens for advanced stage HL used extensively in Europe include escalated BEACOPP (bleomycin, etoposide, doxorubicin, cyclophosphamide, vincristine, procarbazine, prednisone) and ABVD, along with BV-AVD (brentuximab vedotin, doxorubicin, vinblastine, dacarbazine) in recent years [12][13][14]. Slovenian guidelines for treatment of malignant lymphomas propose treatment with ABVD for early stages of HL and ABVD or escalated BEACOPP for advanced stages HL [15].
Intensi ed rst-line chemotherapeutic regimens (escalated BEACOPP) have been designed to overcome the risk of early chemo-resistance development. However, the treatment-related toxicity of intensive approaches is fairly high and is associated with complications that could delay the administration of further chemotherapeutic cycles or could lead to the cytostatic dose reductions [16]. Relative dose intensity (RDI) represents the ratio of the amount of a drug actually administered to the patient in regard to the amount planned for a xed time period [17,18]. The purpose of calculating the RDI is to evaluate whether or not the planned dose intensity of a chemotherapy treatment was actually achieved.
The aim of this retrospective study was to investigate the association of RDI with the outcome of HL patients with advanced stage disease receiving ABVD and escalated BEACOPP regimen.

Patients
We retrospectively reviewed medical records of histologically con rmed HL patients at the Institute of Oncology Ljubljana, Slovenia, between 2004 and 2013. We enrolled patients with advanced stage disease that were planned to receive either 8 cycles of ABVD or 4 cycles of escalated and 4 cycles of regular BEACOPP until 2011 and, from 2012 on, 6 cycles of escalated BEACOPP, because of modi cation of national guidelines. In line with this guidelines, ABVD was reserved for patients older than 60 years and for younger patients, who were un t to receive escalated BEACOPP (patients with severe arterial hypertension, chronic obstructive pulmonary disease, and diabetes mellitus) or reluctant for aggressive chemotherapy (childcare, work during treatment) [15].
Before the initial treatment, physical and blood examination, echocardiography, chest X-ray, computed tomography or positron emission tomography-computed tomography, and bone marrow biopsy was conducted in all patients. In addition, the Charlson comorbidity index (CCI), which predicts the risk of mortality associated with a range comorbid conditions, was assessed as well [19].

Relative dose intensity calculation
According to the chemotherapy regimen, all patients were planned to receive the full doses of cytostatic drugs, however, treatment delays and/or dose reductions were found in most of them. RDI was calculated as described below.
Dose intensity (DI), which can be presented as the amount of a drug administered per time unit, is used to assess the intensity of chemotherapy. It is calculated as a dose of a drug per cycle (mg/m2) divided by the planned number of weeks in a cycle [17]. RDI of each drug is acquired as a fraction of actual DI and planned DI, whereas average RDI of chemotherapy regimen as a sum of RDI of each drug divided with the number of drugs in a regimen (4 for ABVD and 6 for BEACOPP). The purpose of calculating RDI of chemotherapy regimen is to evaluate if the planned DI was actually achieved.

Statistical analysis
Categorical variables were summarized with frequencies and percentages, numerical variables with medians, interquartile ranges (IQR) and ranges (due to the asymmetric shape of distributions). Patients' characteristics were compared between treatment groups by using chi-square tests for categorical variables and Mann-Whitney U tests for numerical variables (Table 1).
Overall survival (OS) and progression-free survival (PFS) probabilities were estimated from the end of treatment with Kaplan-Meier method [20]. Time to second malignancy was not analysed due to too few events. The association of variables with OS and PFS was analysed using univariate and multivariate Cox proportional hazards (CPH) models [21]. The proportional hazards assumption was tested using Schoenfeld residuals, and it has not been violated for any of the variables in any of the models.
The main model for OS of ABVD patients was multivariate CPH model with variables RDI and age, allowing only linear effects ( Table 3). As a part of sensitivity analysis, RDI was included also nonlinearly (using restricted cubic splines with three knots). The results remained unchanged, RDI was not statistically signi cantly associated with OS when controlled for age. The association of RDI and age was evaluated with Pearson correlation coe cient. BEACOPP treatment group was not analysed due to too few events. The low number of events per variable prevented also analysis in all patients as groups signi cantly differed in age and RDI which would require too many variables in the model. To demonstrate the strong effect of age on OS, the association of other variables with OS was additionally tested with and without age in the model (Table 4). All analyses were repeated also for PFS, results were similar as for OS (see also Supplementary Figure 1, and Tables S1 and S2).
A p value less than 0.05 was considered as statistically signi cant. All analyses were performed using R statistical software, version 3.6.3 [22].

Patients' characteristics
Between May 2004 and December 2013, 114 patients received treatment for advanced HL and were enrolled for evaluation. Patients' and their disease characteristics are presented in Table 1. The median age of patients was 39 years, majority of patients were males and had stage IV disease. Patients in the BEACOPP group were signi cantly younger, with less comorbidities, and they received a higher RDI (all p<0.001, Table 1).
Dose de-escalation for the escalated BEACOPP chemotherapy follows a prede ned scheme, which is determined by the occurrence of toxic events in the previous cycles, such as leukopenia, thrombocytopenia and other toxicities [23]. Treatment always begins at dose level 4, which is later reduced as necessary to level 1, before regular BEACOPP is used. In BEACOPP group, the majority of RDI reductions was a consequence of reduced cytostatic doses according to de-escalation protocol, whereas in ABVD it was mostly caused by nonprotocol dose reductions and treatment delays. Overall survival and PFS were signi cantly better in the BEACOPP group compared to ABVD group (both p<0.001, Figure 1, Supplementary Figure S1). However, direct comparison between the two groups is not reasonable, because the groups differed markedly, especially in terms of age and CCI. Median age in the ABVD group was 59.8 years, whereas it was only 32.9 years in the BEACOPP group. Furthermore, 27 patients in ABVD group were older than 60 years while none was in the BEACOPP group. Likewise, the median CCI was 4 in ABVD and 2 in BEACOPP group, respectively. Twenty-eight % of patients in ABVD group and only 2% in BEACOPP group had a CCI of more than 5, for which the estimated 10-year survival is 2% or lower [19]. Fig. 1 Overall survival of all Hodgkin lymphoma patients (a) and for ABVD (doxorubicin, bleomycin, vinblastine, dacarbazine) and BEACOPP (bleomycin, etoposide, doxorubicin, cyclophosphamide, vincristine, procarbazine, prednisone) treatment groups separately (b); shaded areas represent 95% con dence intervals, censoring times are marked with crosses The different patients' characteristics in treatment groups led to a separate analysis of the ABVD group. The low number of events prevented further analysis in the BEACOPP group. In ABVD group, RDI was not signi cantly associated with OS (p=0.590) or PFS (p=0.354) in a multivariate model where age was controlled (see Table 3 for OS and Supplementary Table S1 for PFS). As a part of sensitivity analysis, we included RDI in models for OS and PFS also nonlinearly, the effect of RDI remained non-signi cant (p=0.436 for OS, p=0.434 for PFS). This could be explained with a strong negative correlation between the RDI and age (Figure 2). Pearson correlation coe cient was -0.61 for all patients and -0.45 for ABVD treatment group, indicating that patients with higher age received a lower RDI.

Discussion
The ABVD and the BEACOPP represent the standard of care for patients with advanced HL [13,24]. The aim of this study was to assess the association of RDI with the outcome of HL patients receiving either ABVD or BEACOPP regimen. Multiple works have reported an association between the RDI and survival prognosis, especially in breast cancer and aggressive lymphoma [25][26][27][28]. However, only a few studies have addressed this issue in HL.
The results of present analysis indicate that there is no evidence that a higher RDI results in better prognosis of HL patients. These ndings are in concordance with the key study published on this topic on 380 patients by Owadally and colleagues, who also found no clear evidence that DI in uences the outcome [29]. However, in the study of Owadally and colleagues the DI was measured only in rst two cycles of ABVD chemotherapy, whereas in our study the RDI was measured throughout the whole treatment with either ABVD or BEACOPP, the treatments lasting from 6 to 8 cycles. It is worth noting that it is especially hard to maintain high DI in the last cycles of treatment, particularly on account of accumulated toxicities and complications from previous cycles. The median RDI in the study of Owadally and colleagues was 89% with the lower and upper quartiles of 79% and 97%, respectively, whereas we found a median RDI of 82.3% and the lower and upper quartiles of 68.2% and 89.9% for the ABVD group, respectively. Patients included by Owadally and colleagues were younger, with a median age of 36 years, having both early and advanced stages of HL, while our patients in the ABVD group were characterized with advanced stage disease and a median age of 59.8 years. Therefore, the calculation of RDI for all 8 cycles of ABVD, the older age of patients and the advanced stage HL in all patients might explain the difference in lower RDI achieved in our study.
Similar conclusions were also drawn by Raida and colleagues [30]. Likewise, they found no in uence of primary chemotherapy DI on the probability of complete remission, disease relapse, event-free survival and OS. They included 194 heterogeneous patients with predominantly early HL (63.4%), who had the median age of 28 years and have received diverse chemotherapeutic regimes, including ABVD, BEACOPP, and Stanford V among others. In the study of Raida and colleagues, the median RDI was not reported, however 76.3% of patients received a RDI of 90% or more, which is considerably higher than 51.8% of patients with the RDI of 90% or more achieved in our study for both chemotherapy groups combined. Again, the reason for the difference in the attained RDI is most likely due to the signi cantly older patient population and more intensive chemotherapy regimens in our study. As shown in Results, because of a strong negative correlation between the RDI and age, we can assume that patients with higher age achieve a lower RDI.
Landgren and colleagues evaluated the effect of RDI on prognosis of 88 elderly (> 60 years) HL patients, though the nal study cohort consisted of 59 patients only [31]. Unlike our study and previous two studies, Landgren and colleagues reported a signi cantly better OS in patients with the RDI > 65% compared to those with the RDI ≤ 65%, despite the relatively low number of enrolled patients. It is worth noting that RDI was not controlled for age in a multivariate model, which might explain the association with OS. Similar to the report of Owadally and colleagues, the calculated RDI values were based on the initial two cycles of chemotherapy only. Patients included by Landgren and colleagues had various stages of HL, the majority (69.3%) of them being advanced stage, they also received ve different chemotherapy protocols. We agree with Raida and colleagues that the RDI of ≤ 65% suggested by Landgren and colleagues may be considered as a signi cant violation of primary chemotherapy protocol, and is as such not appropriate to arbitrary divide the patients with good or bad prognosis.
Our study has several limitations that merit consideration. The major limitation is the modest sample size and the difference in patients' characteristics between the ABVD and BEACOPP groups. The ABVD group was considerably older and had more comorbidities, therefore we cannot compare the two groups directly. Additionally, the RDI analyses performed in our study were retrospective in design, therefore only hypotheses about possible association can be formulated.
The available evidence suggests that small dose reductions or short delays between chemotherapy cycles, which still result in a decreased RDI, may not affect overall outcomes of HL patients, most likely due to a relatively good prognosis and chemosensitivity of this disease. To our knowledge, this is the rst study to evaluate the impact of RDI throughout whole treatment in patients with advanced HL treated exclusively with ABVD or BEACOPP chemotherapy. The lack of association between the RDI and response to treatment is in concordance with the current literature. However, in order to fully elucidate the relationship between the RDI and response, a prospective trial with a larger number of patients would be required.

Declarations Author contributions
Rožman Samo contributed to the conception of the study. Rožman Samo and Jezeršek Novaković Barbara designed the study and analysed the clinical data. Ružič Gorenjec Nina performed statistical analysis. All authors contributed to the acquisition, analysis or interpretation of data for this article and drafts of the article. All authors participated in writing the manuscript and approved the nal version of it. All authors were involved in revising the paper critically for intellectual content, and gave nal approval for the submission of the paper.

Funding
This research work was partially supported by a grant from Slovenian Research Agency (Program P3-0321).

Data availability statement
The data that support the ndings of this study are available from the corresponding author on reasonable request.

Ethics approval
The study was approved by the Institutional Review Board at the Institute of Oncology Ljubljana (Approval number KSOPKR/72) and National Medical Ethics Committee of Republic of Slovenia (Approval number 0120-481/2017/5).