Prophylaxis with trimethoprim-sulfamethoxazole (TMP-SMX) decreases the incidence of interstitial pneumonia in B cell non- Hodgkin lymphoma patients receiving chemotherapy with rituximab CURRENT STATUS: UNDER REVISION

Several studies have reported incidence of interstitial pneumonia (IP) among patients with non-hodgkin lymphoma (NHL) receiving chemotherapy with rituximab, however the best prophylaxis method remains unclear. This retrospective study was intended to identify the effect of prophylactic TMP-SMX once daily on IP and high risk factors effecting IP in patients with NHL. From March 2013 to April 2018, 498 patients (264 male, 53%) with B cell NHL receiving first-line CHOP-like chemotherapy with rituximab were enrolled in this study. The median age of the patients was 56 years old. Among these patients, 311 patients (62.4%) received prophylaxis of TMP-SMX. IP occurred in 65 patients, so the incidence rate of IP was 13.1%. The prophylactic use of TMP-SMX given one tablet daily could significantly decrease the rate of IP from 21.4% to 8.0% (p<0.001). For 311 patients with prophylactic TMP-SMX, 2 (1.2%) had rashes, 38 (12.2%) suffered nausea and vomiting, 52 (16.7%) exhibited neutropenia, and 18 (5.8%) suffered kidney dysfunction. Being male, having a history of diabetes and not receiving TMP-SMX prophylactic therapy were considered as statistically significant risk factors for IP in univariate and multivariate analysis. Disease progression was observed in 55/311 (17.7%) patients with prophylactic TMP-SMX and in 63/187 (33.7%) patients without prophylactic TMP-SMX treatment (p<0.001). The present study concluded that IP occurrence is not rare in B cell non-hodgkin lymphoma patients receiving chemotherapy with rituximab, while prophylaxis of oral TMP-SMX given one tablet daily significantly decreased incidence of IP.

effective prevention of IP is important for patients undergoing immunochemotherapy. Rituximab combined with chemotherapy could inhibit immune function of patients, and increase opportunistic infection. Several studies concluded elevated the incidence of IP possibly because of an increase in the risk of PCP 4-6,10,11 . However, the most effective prophylactic drugs and methods for preventing IP remain controversial. Trimethoprim-sulfamethoxazole (TMP-SMX), an oral broad-spectrum antibiotic, is a specific therapeutic and prophylactic agent used to treat pneumocystis carinii pneumonia (PCP) infection 5,6,11 . Since 2013, hundreds of patients with lymphoma have taken prophylaxis of TMP-SMX in our hospital. Here, we present this retrospective study summarizing the effects of TMP-SMX prophylaxis as a means of reducing IP incidence among patients of B cell non-Hodgkin lymphoma via a combination of chemotherapy and rituximab.

Participants and data collection
We retrospectively reviewed treatment data from patients with CD20+ B cell non-Hodgkin lymphoma as diagnosed by a local pathologist according to WHO criteria. 12 All patients had received rituximab combined with CHOP-like chemotherapy for at least two cycles. Prophylactic granulocyte colonystimulating factor (G-CSF) and antiemetic administration was allowed. Prophylactic TMP-SMX was taken once daily from treatment initiation to completion of chemotherapy. Each tablet contained 0.08 g TMP and 0.4 g SMX. TMP-SMX use was not permitted in patients with allergies to the drug, megaloblastic anemia, or severe liver or kidney damage. A group of patients that did not undergo prophylactic TMP-SMX treatment were analyzed as control group. Radiographic documents and data regarding patient clinical characteristics, histological diagnosis, chemotherapy regimens, and survival outcome were extracted from the original computer medical files. The trial was approved by the research ethics boards of Zhejiang cancer hospital, with all patients giving written informed consent.

IP Diagnosis and treatment
The observation period for IP extends from the first day of immunochemotherapy to 8 months after completion of immunochemotherapy. Routine imaging evaluations were performed every two cycles during chemotherapy and every 3 months after chemotherapy in observation period. Patients received thoracic CT scans when they exhibited symptoms of pulmonary infection. Diagnosis of IP can be done via multidisciplinary approach in which the clinical symptoms, laboratory tests, radiologic and pathologic findings. Typical imaging presentations of IP was diffuse pulmonary interstitial infiltrates with reticular or ground-glass opacity, alveolitis, and the presence of diffuse infiltrates on CT scans. 11,13 When IP was suspected, laboratory tests measuring blood cell counts, C reactive protein (CRP), procalcitonin (PCT), 3-b-D glucan antigens, Gram-negative lipopolysaccharides, Galactomannan antigen detection (GM), and bacterial culture were conducted to facilitate appropriate diagnosis. Once diagnosed with IP, patients started empirical treatment with a combination of antibiotics, antifungal agents, and glucocorticoids. Ganciclovir was given when virus infection was suspected. TMP-SMX should be used at therapeutic dose if PCP was suspected. CT scans were conducted weekly until complete interstitial infiltrate absorbance was evident. After patients recovered from IP, retreatment with chemotherapy and/or rituximab was allowed. TMP-SMX was administered at prophylactic doses as described previously. Parameters relating to clinical presentation, diagnosis, causative pathogen, treatments, and outcomes were additionally summarized for all patients as appropriate.

Statistical analysis
For this study, overall survival (OS) and progression-free survival (PFS) were measured, with the former corresponding to the amount of time between starting chemotherapy and last follow-up or death of any cause, and the latter corresponding to the period of time from start of chemotherapy to last follow-up, cancer progression, or all-cause death. The National Cancer Institute Common Terminology Criteria for Adverse Events (NCI-CTCAE) v 4.0 criteria were used for gradation of all adverse events (AEs). χ2 tests were used for comparing categorical variables between patient groups.
A binary logistic regression was used to conduct univariate analyses with appropriate hazard ratios (HRs) with 95% confidence intervals (CIs). Each variable with a p<0.05 in the initial univariate analysis was incorporated into a multivariable model. P<0.05 was the threshold of significance. SPSS 23.0 was used for all statistical testing.
Diagnosis and treatment of IP IP occurred in 65 patients (13.1%), of whom 25 (38.5%) were in the TMP-SMX prophylaxis group. IP patients were a median of 60 years old (range: 18 -78), with the following pathological diagnoses: 57 cases of DLBCL, 1 of FL, 1 of MCL, 1 of CLL/SLL, 3 of MZL, and 2 of "highly aggressive B cell lymphoma". IP incidence was significantly greater among patients who did not receive TMP-SMX prophylaxis relative to those who did (21.4% vs 8.0%, p<0.001). The median time between the first day of immunochemotherapy and the date of IP diagnosis was 74 days, (range: 7 -158 days). Prior to IP diagnosis, patients had been treated with a median of 3 rituximab-chemotherapy cycles (range: 1 -8 cycles). The median cumulative dose of rituximab upon IP diagnosis was 2100 mg (range: 600 -4800 mg). A review of these 65 cases is shown in Table 2.
All patients diagnosed with IP were assessed for measurements of inflammatory markers. Elevated CRP (> 10mg/L) was evident in 54 patients. Elevated PCT (>0.5ng/ml) was evident in 5 patients. A positive G test for 3-b-D glucan fungal antigens (> 60 pg/mL) was detected in 18 patients. Elevated levels of Gram-negative lipopolysaccharides (> 10pg/mL) were detected in 36 cases. Hypoxemia was detectable in 5 cases as determined based upon arterial blood gas measurements. Sputum cultures were performed in patients following expectoration, and positive pathogenic bacteria included Candida albicans (n=3), Staphylococcus aureus (n=1), Klebsiella aeruginosa (n=2), Klebsiella pneumonia (n=2), hemolytic Steptotoccus (n=1), and Staphylococcus haemolyticus (n=1). Blood culture was performed in pyrexic patients with temperatures higher than 38.5 °C, but the results were all negative for all 8 tested patients. 7 patients received bronchoalveolar lavage (BAL) but no Pneumocystis carinii was detected. Other common suspicious pathogens included haemophilus, stenotrophomonas, tropheryma, exophiala, human gammaherpesvirus 4.
Further chemotherapy and rituximab treatments were withheld immediately following IP diagnosis.
The average time to IP remission was 12 days (range: 7 day to 58 days). No patients died as a consequence of their infection. After remission from IP, 21 patients completed further chemotherapy combined with rituximab, 35 patients received subsequent chemotherapy without rituximab, and 9 patients did not receive chemotherapy nor rituximab. No patients suffered a recurrence of IP following continued immunochemotherapy treatments.

Complications of prophylaxis
Of the 311 patients receiving prophylactic TMP-SMX, 2 (1.2%) had rashes, 38 (12.2%) suffered nausea and vomiting, 52 (16.7%) exhibited neutropenia, and 18 (5.8%) suffered kidney dysfunction. No patients discontinued TMP-SMX prophylaxis as a result of adverse reactions. Because of concomitant chemotherapy, it is difficult to determine whether these adverse reactions were specifically associated with TMP-SMX prophylaxis in most cases.

IP Risk factors
To assess IP risk factors in study participants, we examined different clinical parameters (Table 3). In a univariate analysis, being male, having a history of diabetes, and not receiving TMP-SMX prophylactic therapy were identified as IP risk factors. In a subsequent multivariate analysis, all three of these variables were found to independently predict a higher IP risk.

Impact of IP on survival
After a median 26 months follow-up period, median PFS and OS were 23.1 and 26.7 months, respectively, in all patients. Among the 311 patients receiving prophylactic TMP-SMX, 55 exhibited disease progression, while among the 187 patients that did not receive prophylactic TMP-SMX, 63 exhibited disease progression -a significantly larger proportion (17.7% vs 33.7%, p<0.001).

Discussion
As rituximab combined with chemotherapy was widely employed to treat non-Hodgkin lymphoma, there are increasing reports regarding treatment-associated IP. However, there is substantial variability among previous studies with respect to IP incidence among lymphoma patients. In one retrospective study of 2212 consecutive Chinese lymphoma, overall IP incidence was determined to be 3.75%, with 3.9% (7/287) and 2.4% (76/925) in patients with hodgkin and non-Hodgkin lymphoma, respectively 9 . However, in other studies the incidence of IP among non-Hodgkin lymphoma following CHOP-based chemotherapy with or without rituximab has ranged from 1. These studies have also suggested that the addition of rituximab to therapeutic regimens may increase IP incidence. In the present study, we observed an IP incidence of 21.4% in patients not receiving preventive treatment, which is higher than in previous reports. There are several possible reasons for this difference. As an anti-CD20 antibody with an extended in vivo half life, rituximab can modulate many immune processes. It can induce the apoptotic death of B cells, alter complement activation and induce the release of particular cytokines, thereby potentially interfering with normal cytotoxic T cell responses and immune functioning, thereby facilitating the occurrence of various opportunistic infections 18 . In this study, all patients received rituximab and chemotherapeutic treatments simultaneously, which may also explain our results. It is also possible that the IP incidence rate was higher because patients were being actively monitored for this condition, leading to increased diagnostic rates and detection in asymptomatic patients. Other possible causes include differences in baseline characteristics of patient populations, chemotherapy regimens administered, chemotherapy dose intensity, diagnostic techniques, the small size of previous studies, or the fact that we employed a longer observation period. Therefore, our results suggest that it will be necessary to administer prophylactic IP treatment to patients undergoing therapy as in this study.
The pathological changes of IP refer to the inflammatory changes in the interstitial tissue, mainly the supporting tissues outside the alveolar and terminal epithelial cells, including blood vessels and untreated patients suffering a 21% PCP incidence and treated patients suffering a 0% incidence rate both if TMP-SMX was administered daily or 3 days per week 26,27 . More recently, many studies have also confirmed the efficiency of TMP-SMX prophylaxis as a means of decreasing the incidence of PCP in adult patients with lymphoma or in pediatric oncology patients 15,[28][29][30][31] . A meta-analysis identified twelve randomized trials and found that TMP-SMX administration was linked to a 91% drop in the incidence of PCP, with a significant reduction in PCP-related mortality 23 . Consistent with these findings, in the present study we found that prophylactic TMP-SMX treatment significantly decreased the incidence IP in B cell lymphoma patients undergoing R-CHOP-like chemotherapy from 21.4% to 8.0% (p<0.001).
The optimal administration schedule for prophylactic TMP-SMX is not well defined. Most previous studies have concluded that intermittent dosing with TMP-SMX is an effective alternative prophylactic regimen. TMP-SMX is thus often given twice daily two times a week 30 , 2 consecutive days per week 28,32,33 , twice weekly 31 , or three days a week 27 . Intermittent TMP/SMZ is effective for preventing PCP, and intermittent dosing is linked to reduced costs and a lower rate of fungal infections. However, a study by Toshiro et al. instead gave one TMP-SMX tablet daily throughout chemotherapy 15 . A metaanalysis concluded that lower doses of TMP-SMX were an effective means of improving tolerance while not compromising primary prophylactic efficacy 34 . No differences between once-daily and thrice-weekly administration schedules have been found 23 . TMP-SMX was given as one tablet daily in the present study, and this approach was convenient and easy to implement.
Roughly 30% -40% of patients stop TMP-SMX therapy as a result of poor drug tolerance when receiving intermittent prophylactic treatment, according to previous reports 35 . The most common AEs linked with such discontinuation include skin rash, myelosuppression, nausea, fever, renal and liver toxicity, and hyperkalemia 36-39 . The observed AEs associated with prophylactic TMP-SMX in the present study were consistent with these previous reports, however drug tolerance for this daily TMP-SMX regimen was high, with no discontinuations due to adverse reactions. Moreover, the incidence of leukopenia and nausea and vomiting was low, which may be explained by the fact that patients were allowed to receive prophylactic G-CSF injections after chemotherapy and antiemetic treatments in this study.
We found that a history of diabetes, being male, and not undergoing prophylactic TMP-SMX treatment were independent risk factors for IP. In diabetic patients, hyperglycemia can affect the chemotaxis, adhesion, phagocytosis and intracellular bactericidal efficacy of immune cells. In addition, the thickening of alveolar epithelium, vascular hyaline degeneration, and pulmonary microangiopathy in diabetic patients can affect lung function. These factors will damage immune function and increase rates of opportunistic infections among patients with diabetes, who experience 30% more pneumoniarelated mortality than do non-diabetic patient populations 40 . Males usually receive higher doses of rituximab, have a longer smoking history, and are more likely to have poorer basic lung function than are females. Other factors from previous studies included application of rituximab 13-15,17,41 , pretreatment absolute lymphocyte counts <1x10 9 /L 17 , B symptoms (fever, weight loss, night sweat), a drug allergy history 9,14 , and increased intensity of corticosteroid exposure 5,42,43 .
Owing to the following limitations, caution is warranted when interpreting the results of this study. For one, BAL, as an important auxiliary examination procedure, has not been widely used in this study.
Only 7 patients received BAL and no PCP infection was found. We therefore lack sufficient evidence regarding the specific cause of IP. Secondly, there were still 8% patients developed IP after prophylactic treatment of TMP/SMX, which meant there may be other causes. Therefore, more prospective studies are needed to explore other prophylactic drugs and their optimal administration.
In addition, the retrospective nature of this study implies a potential information bias, potentially explaining the observed discrepancies regarding rates of side effects associated with TMP-SMX.

Conclusions
It is clear that among B cell non-Hodgkin lymphoma patients undergoing chemotherapy with rituximab, IP is a relatively common occurrence. We conclude that prophylaxis with once daily oral TMP-SMX from the start of immunochemotherapy to the completion of chemotherapy significantly decreases the incidence of IP.