Clinical characteristics and outcomes for HIV-associated Diffuse Large B-cell Lymphoma in China: A retrospective single-center study

Background: Human immunodeciency virus (HIV) infection is associated with an increased risk of aggressive lymphoma, especially diffuse B cell lymphoma (DLBCL). We aimed to analyze characteristics and outcomes of DLBCL in HIV-associated patients. Methods: We retrospectively studied HIV-infected patients with DLBCL since 2011. Data on HIV infection and lymphoma characteristics, treatment and outcome were retrieved and analyzed. Results: In 53 patients with HIV-associated DLBCL, most patients had frequent bad performance status (PS) (74%), elevated LDH level (94%), B symptoms (69%), advanced Ann Arbor (75%), bulky tumor (72%) and extra-nodal involvement (68%) at diagnosis. The median CD4 T cell count was 175/μl, and 17 patients were already on cART treatment. Plasma EBV DNA was detectable in 18 patients (53%, 18/34). Of the patients evaluated at the end of treatment, 21 (64%) achieved CR, 1 (3%) achieved PR and 6 (18%) experienced progressive disease. The 2-year progression free survival (PFS) was 50% and overall survival (OS) rate was 60%. Factors associated with poor PFS in univariate analysis were unfavorable PS, high IPI score and detectable plasma EBV load. High IPI and detectable EBV load correlated with worse OS. However, only detectable plasma EBV load was the independent factor of poor prognosis for PFS (HR: 14.46, 95% CI [2.57-81.50]) and OS (HR: 7.47, 95% CI [1.48-37.59]) in a multivariate Cox regression model. Conclusions: In our population, HIV-associated DLBCL still presented aggressive characteristics and poor survival outcomes. Plasma EBV DNA could be used as a prognostic factor in HIV-infected DLBCL. The extra-nodal the gut (n 19), bone (n (n (n (n 4) (n One-third system-IPI (CNS-IPI) at Among 34 cases on EBV elevated (5×10 in 18 (53%). Of all patients, six patients (11%) presented with positive HBsAg and four (7.5%) presented with positive anti-HCV antibody. methotrexate and cytarabine for central nervous system prophylaxis. Radiotherapy with 40–50 Gy was given in ten patients with bulky tumor(cid:0)among which 4 patients received radiotherapy as consolidation after chemotherapy with CR and 6 patients were treated with concurrent chemoradiotherapy. Three patients were given autologous hematopoietic stem cell transplant (auto-SCT) following chemotherapy.


Introduction
Human immunode ciency virus (HIV) infection is associated with an increased incidence of malignancy. HIV-associated diffuse large B cell lymphoma (DLBCL) and Burkitt's lymphoma (BL) are among the commonest cancers in HIV-positive patients [1][2][3]. Other non-Hodgkin's lymphomas (NHLs), including primary effusion lymphoma (PEL), plasmablastic lymphoma, KSHV-associated multicentric Castleman's disease, primary central nervous system (CNS) lymphoma and classic Hodgkin's lymphoma (cHL), are diagnosed in patients with HIV as well [1]. It has been previously studied that HIV strongly contributed to the NHL mortality, particular in acquired immunode ciency syndrome (AIDS)-de ning subtypes [4]. With the widespread administration of combination antiretroviral therapy (cART), the incidence of HIV-associated lymphoma has decreased and an improvement in the survival outcome has been observed [4,5].
DLBCL is the most common subtype of lymphoid malignancy in adults. It represents approximately 30% of NHL cases without HIV infection and 45% of cases with HIV-related lymphoma [6][7][8]. In a Spanish study, compared with HIV-uninfected DLBCL patients, HIV-infected patients had more aggressive features with a poorer performance status, more frequent B symptoms and more advanced Ann Arbor stages [5]. In the above study, when treated with the standard-of-care regimen rituximab plus cyclophosphamide, hydroxydaunorubicin, vincristine and prednisolone (R-CHOP), HIV-infected patients had a similar disease-free survival but a signi cantly worse overall survival (OS), compared to those without HIV infection [5]. In contrast, a French study revealed that HIV-infected patients with DLBCL showed survival outcomes similar to those of HIV-negative DLBCL patients [9].
Infections with Epstein-Barr virus (EBV), Hepatitis B virus (HBV), and/or Hepatitis C virus (HCV) have been reported previously in DLBCL, and the former two factors independently predicted poor prognosis [10][11][12][13]. In HIV-infected patients with DLBCL, it has been shown that elevated EBV load predicted inferior survival outcome; however, few data on HIV and HBV/HCV co-infected situation in patients with DLBCL were published [13].
The HIV epidemic appears to have obvious regional distribution in China, occurring frequently in Sichuan province and Yunnan province [14,15]. So far, there are only few studies described the clinical features of Chinese AIDS patients with DLBCL [16,17]. Herein, we retrospectively analyzed the characteristics and outcomes of Chinese HIV-infected DLBCL patients in Yunnan Provincial Infectious Diseases Hospital/Yunnan AIDS Care Center.

Patients
We retrospectively included 53 HIV-infected individuals with newly diagnosed DLBCL between 2011 and 2018 in Yunnan Provincial Infectious Diseases Hospital/Yunnan AIDS Care Center, which is the largest HIV/AIDS referral hospital in southwestern China. This study was approved by institutional review board and conducted according to the declaration of Helsinki. The diagnosis was established in accordance with 2008 WHO classi cation. DLBCL was classi ed by the Hans classi cation algorithm. The clinical data were collected from medical records. Data of patient's demographic characteristics (gender and age), HIV-related characteristics (HIV transmission route, years of HIV infection at DLBCL diagnosis, CD4 cell count at DLBCL diagnosis and date of cART initiation), lymphoma-related characteristics (cell-of-origin subtype, Eastern Cooperative Oncology Group (ECOG) performance status score, serum lactate dehydrogenase (LDH), B symptoms, extra-nodal sites, Ann Arbor stage, international prognostic index (IPI) score, bulky tumor, bone marrow involvement and CNS-IPI score) and other related characteristics (EBV load, HBV load, HCV load, comorbidities, time from rst complain to diagnosis) were available.
In this study, cART regimen included two nucleoside reverse transcriptase inhibitors and one non-nucleoside reverse transcriptase inhibitor. Patients after diagnosis of DLBCL initiated or were switched to the cART treatment with tenofovir, lamivudine, plus efavirenz or raltegravir potassium. Patients with a CD4 + cell count below 200 cells/μl received trimethoprim/sulphamethoxazole against Pneumocystis jiroveci as prophylactic agent. HIV-1 RNA load was monitored during chemotherapy. For patients with positive hepatitis B surface Ag (HBsAg), prophylaxis was not used because both tenofovir and lamivudine have antiviral activity against HBV. HBV DNA (cutoff ≥ 5×10 2 copies/ml) and HCV RNA loads (cutoff ≥ 5×10 2 IU/ml) were assessed during chemotherapy as well.
Laboratory monitoring of plasma EBV DNA (cutoff ≥ 5×10 3 copies/ml) load was performed in some patients.

Response assessment
Interim evaluation was performed after completion of three or four cycles of chemotherapy. After one month of the completion of all the treatment, the e cacy evaluation was conducted. Computed tomography (CT) or 18 F-uorodexyglucose positron emission tomography (PET) was performed for radiological evaluation. Brain Magnetic Resonance Imaging (MRI) was used for CNS involvement evaluation. The 2007 revised Cheson criteria were performed to de ne complete response (CR), partial response (PR), progressive disease (PD) and relapse.

Statistical analysis
All statistical data were analyzed using SPSS software, version 18 or GraphPad Prism 6. Continuous variables were presented as the median with the rst quartile and third quartile, and categorical variables as numbers and percentages. PFS was de ned as time from DLBCL diagnosis to progression, relapse or death from any cause. OS was de ned as time from DLBCL diagnosis to last follow-up or death from any cause. Survival curves were plotted using the Kaplan-Meier method and the log-rank test was used for comparison. Cox proportional regression models were performed for univariate and multivariate analyses of outcomes. P value 0.05 was considered as statistically signi cant.

Patient characteristics
The baseline clinical features of HIV-infected patients were summarized in Table 1. Of these patients, median age was 42 and 46 (87%) of them were male.
The main HIV transmission group was heterosexuals. Among 53 patients, 20 patients had a HIV infection history more than 1 year at the time of DLBCL diagnosis, and in 24 patients, DLBCL and HIV infection were diagnosed concomitantly. The median CD4 T cell count at DLBCL diagnosis was 175/μl (range 26-551) in these 53 patients, of whom 4 patients had a CD4 T cell count less than 50/μl. At DLBCL diagnosis, 17 patients were already on cART treatment. The median time from patients' rst complain to diagnosis was 2 months. Most patients (74%) had a bad performance status (PS) (ECOG PS 2-4), and elevated LDH at diagnosis were present in 94%. Five patients presented with 1 or 2 comorbidities (9%). Six-nine percent of patients had B symptoms, 75% had an advanced Ann Arbor stage ( -) and 72% had bulky tumor at diagnosis. Thirty-six patients (68%) had extra-nodal involvement, of whom thirteen patients with more than 2 extra-nodal sites. The most frequent extra-nodal sites were the gut (n = 19), bone marrow (n = 7), stomach (n = 6), bone (n = 6), liver (n = 4) and/or kidney (n = 4). One-third of the patients had a higher central nervous system-IPI (CNS-IPI) at diagnosis. Among 34 cases with information on EBV status, EBV load was elevated (5×10 3 copies/ml) in 18 (53%). Of all patients, six patients (11%) presented with positive HBsAg and four (7.5%) presented with positive anti-HCV antibody.

DLBCL treatment
Forty-two (42/53) patients received CHOP±R regimen among which twenty-one patients received six to eight cycles CHOP±R regimen. Nine of these 42 patients received R-CHOP and seven received CHOP regimen, while ve received CHOP followed by R-CHOP or vice versa depending on their nancial situation at that time. A total of twelve patients stopped CHOP±R after a median cycle of 3 when symptoms improved without respect to our advices. One patient who received 4 cycles R-CMOP (cyclophosphamide, mitoxantrone, vincristine and prednisone) was switched to R-GOD (gemcitabine, oxaliplatin and dexamethasone) plus oral lenalidomide due to PD. Six patients who received less than 6 cycles CHOP±R regimen died on therapy. Nine patients with high aaIPI or high-grade disease received R-EPOCH/R-DA-EPOCH (etoposide, prednisone, vincristine, cyclophosphamide and doxorubicin) regimen. Two patients received rst-line R-CHOP progressed in interval evaluation and one was changed to GemOx (gemcitabine and oxaliplatin) regimen and the other was switched to GDP (gemcitabine, dexamethasone and carboplatin) regimen. One patient received four cycles R-CHOP was in PR in interval evaluation; however it was switched to R-GDP regimen due to a high SUVmax of interval PET/CT evaluation. Thirty-eight out of 53 patients received intrathecal methotrexate and cytarabine for central nervous system prophylaxis. Radiotherapy with 40-50 Gy was given in ten patients with bulky tumor among which 4 patients received radiotherapy as consolidation after chemotherapy with CR and 6 patients were treated with concurrent chemoradiotherapy. Three patients were given autologous hematopoietic stem cell transplant (auto-SCT) following chemotherapy.

Treatment response and outcomes
Among forty-seven HIV-infected DLBCL patients who are evaluable in interim analysis, the overall response rate (ORR) was 88%, including 22 (47%) CR and 19 (41%) PRs. Three patients experienced PD and three patients died before the interval evaluation. One died from hemorrhage, one from severe pleural effusion due to mycobacteriumtuberculosis and one from rapid progression of disease. Of the 33 patients evaluated at the end of treatment, 21 (64%) achieved CR, 1 (3%) achieved PR and 6 (18%) experienced progressive disease ( Table 2). The other 5 patients died during chemotherapy, before response could be evaluated, from sepsis (n = 4), or heart failure (n = 1). No one included in our cohort had baseline CNS involvement. With over 70% patients received CNS prophylaxis, one patient had a CNS relapse 4 months after reaching CR, died quickly.
The median PFS for HIV-infected DLBCL patients in our study was 24 months and median OS was not reached. The overall 2-year PFS and OS rates were 50% (95% CI [33%-65%]) and 60% (95% CI [42%-74%]), respectively ( Fig. 1A (Table 3). No HIV related characteristics were associated with PFS and OS. Neither HBV nor HCV coinfection was associated with prognosis. The survival curve of plasma EBV load positive patients (n = 18) and plasma EBV load negative patients (n = 16) for PFS and OS were shown in Figure 1C and 1D, reaching signi cantly statistical difference (p = 0.001 and p = 0.006).

Discussion
Yunnan is located in southwest China, border with Myanmar, Laos and Vietnam, and has been regarded as the most major site of HIV infection epidemic for a long time [14,15]. Over the decades, people living with HIV/AIDS is increasing in Yunnan, and it is still one of the provinces with highest number of HIVinfected patients in China [18].
DLBCL is the most common pathological subtype of NHL in both the general population and people living with HIV/AIDS [1][2][3]19]. A large database study shows that HIV infection continues to be an independent risk factor for death among patients with lymphoma [20]. We analyzed the clinical features and survival outcomes of HIV-infected DLBCL in the setting of cART in our series. Consistent with other recent studies, HIV-infected patients with DLBCL presented with a poor performance status, a high frequency of B symptoms, elevated LDH, advanced Ann Arbor stage and high aaIPI score [5,6,9,17,20]. Moreover, bulky disease was frequent. EBV infection is found in approximately 10% of HIV negative DLBCL [21]; however, in our study, 53% of patients with HIV positive DLBCL had elevated plasma EBV DNA load, suggesting a higher rate of EBV co-infection. In contrast, the frequency of HBV infection of HIV-infected patients with DLBCL was similar with their HIV-uninfected counterparts reported in Chinese population [11]. Likewise, the frequency of HCV infection was close to that of HIV negative DLBCL in Caucasian population but lower than that in Asian population [22,23]. Unlike previous study, in which HIV had been diagnosed a median of 15 years previously and nearly 80% had received cART at DLBCL diagnosis [9], patients in our study revealed the late HIV detection and late cART exposure. It could partially be attributed to the less awareness of HIV testing and less initiative for medical care assistance in high risk HIV population.
Before the introduction of anti-CD20 antibody rituximab, HIV-infected patients with DLBCL treated with CHOP therapy showed a similar response rate and survival as those of HIV-uninfected counterparts [24,25]. Several retrospective studies of DLBCL patients with HIV infection treated with R-CHOP have been reported. In one study from Spain, HIV-infected DLBCL patients had a CR rate of 69% and a 5-year OS rate of 56%, similar to those of 81% and 74% of HIVuninfected patients [5]. Similarly, in a report from France, the survival outcomes of HIV-infected DLBCL patients . In our study, three patients underwent auto-SCT after CR were lymphoma free. At the end of treatment, 64% of evaluable patients achieved CR. The 2-year PFS was 50% and OS rate was 60%. The response rate and survival outcome appear to be lower than others, only similar with a two-year OS rate of 63% from a German cohort, in which mostly patients treated with CHOP-like regimen [7]. One of the reasons might be the less use of rituximab. Medical insurance has not covered the cost of rituximab until the recent years. Besides, Yunnan is a remote province and relatively economically backward, thus, some patients could not afford too much in their treatment and a few had low compliance, did not adherent to the recommended courses of treatment. As a result, the high rate of loss to follow-up (16.9%) made our results less solid.
CNS involvement has been recognized to be more common in AIDS-related lymphomas [28]. A retrospective review of database from clinical trials presented that CNS involvement at baseline was not associated with shortened overall survival, but CNS relapse was associated with a short median OS of 1.6 months [28]. Similarly, one patient in our study had a CNS relapse and died quickly. Although the introduction of prognostic model which is used for assessing the risk of CNS disease in DLBCL has guided the CNS prophylaxis in general DLBCL population, the value of this prognostic model in HIV-infected patients of DLBCL remains unclear. Survival difference was not observed with this prognostic model in our data as most patients received intrathecal prophylaxis, suggesting the role of routine CNS prophylaxis in overcoming the unfavorable prognosis of high CNS-IPI.
It has been shown that HIV-related factors, such as a low CD4 cell count and prior history of AIDS are no longer predictive of worse survival outcome in the cART era [29,30], which is consistent with our results. On the other hand, the levels of HIV-1 RNA load were negative in all the subjects during the chemotherapy. We found that unfavorable performance status, high IPI score and elevated EBV DNA load were associated with poor outcomes. Nevertheless, in multivariable analysis, only elevated plasma EBV DNA load was associated with poor outcomes.
It has been widely accepted that EBV plays an important role in the pathogenesis of lymphomas, such as BL, cHL, DLBCL and natural killer (NK)/T-cell lymphoma [31]. In the setting of HIV infection, EBV infection are more frequent, positive in 30-60% of cases compared with about 10% in the general DLBCL patients [6,13,32,33]. We found that of 53% of our DLBCL cases were positive for plasma EBV DNA, consistent with previously reported prevalence of EBVload in HIV associated DLBCL [13]. Its high frequency makes EBV a possible factor which contributes to the development of HIV associated lymphomas and be responsible for driving a more aggressive behavior [32]. In contrast to the fact that blood EBV DNA is not predictive of outcomes in HIV-associated HL [34], previous studies have demonstrated that tumor EBV infection status is an independent adverse predictive factor of survival among patients with HIV-infected DLBCL [35,36]. Meanwhile, Muncunill et al. observed that plasma EBV-load had a negative prognostic impact, and could be used as an early predictor of HIVrelated lymphoma [13]. In this study we showed plasma EBV positivity was an independent negative predictor for both PFS and OS. To our knowledge, this is one of the few studies reported with the prognostic value of plasma EBV DNA in HIV-infected DLBCL, suggesting a possibility of measuring the level of plasma EBV DNA for risk strati cation.
In HIV negative DLBCL, HBsAg-positive patients have worse clinical features and poor outcomes and the presence of HCV revealed inferior OS when companied with impaired liver function [11,37]; whereas our results indicated HIV and HBV/HCV co-infected situation in DLBCL did not predict outcomes. Further studies with more samples needed to re-evaluation.
This study provides important real-world data on outcomes of HIV-infected DLBCL patients in China. A limitation of this study is the small sample size, although this is the largest series reported in China. The survival outcomes remain poor and more therapeutic approaches are warranted. Of note, EBV DNA loads at diagnosis have prognostic value and the use of EBV DNA loads in monitoring disease in patients after treatment needs to be further evaluated.