Weighing the risks and benefits of PJP prophylaxis in rituximab users

Severe infections are a leading cause of morbidity, hospitalization, and mortality for patients with rheumatic diseases. In a recent clinical trial in ANCA-associated vasculitis (AAV), for instance, over 20% of participants had at least one serious infection over 44 months (1) and, in a recent giant cell arteritis (GCA) trial, up to 12% of participants had a serious infection over 12 months (2). Several factors influence the risk of severe infection, including patient age, the immunologic target(s) of the treatment, the presence of end-organ damage related to the rheumatic disease (e.g., lung or kidney disease), and the patient’s comorbidities.

Preventing Pneumocystis jirovecii pneumonia (PJP) is of particular interest for rheumatologists. PJP is often severe among patients with rheumatic diseases, with mortality rates exceeding 40% (3). Prophylaxis with trimethoprim-sulfamethoxazole (TMP/SMX) has been associated with reduced mortality in this population (4, 5). Acknowledging this, the recent American College of Rheumatology / Vasculitis Foundation Guideline for the Management of AAV conditionally recommends prophylaxis to prevent PJP in patients receiving rituximab or cyclophosphamide (6).

Despite the high rate of mortality from PJP and apparent consensus among vasculitis experts, controversy over the appropriateness of PJP prophylaxis persists. The aforementioned recommendations for AAV were not informed by contemporary epidemiologic data, which suggest a lower risk of PJP than previously reported (7–9). In other diseases often treated with high doses of glucocorticoids and other immunomodulators, such as lupus and giant cell arteritis, guidelines have generally not addressed PJP prophylaxis (10, 11). Though generally well tolerated, prophylaxis has been associated with medication-related adverse events, which makes weighing risk and benefits complicated among those at low risk for PJP (7, 12). Finally, glucocorticoid exposure is a strong risk factor for PJP, and recent trends in the use of glucocorticoids have emphasized lower doses for shorter periods than in the past. In the context of these evolving recommendations and uncertainties, studies rigorously investigating the risks and benefits of PJP prophylaxis are needed.

In this issue of Arthritis & Rheumatology, Park and colleagues report the results of a retrospective observational study evaluating the risks and benefits of PJP prophylaxis with TMP/SMX in 818 patients receiving rituximab between 2004 and 2020 for pemphigus, AAV, rheumatoid arthritis, and other rheumatologic indications. Based on an observed 11 infections, the cumulative incidence of PJP at one year was estimated to be 2 cases per 100 person years of exposure among those who did not get prophylaxis and 1 case per 100 person years among those who received prophylaxis. PJP infection was highly morbid: 83% required mechanical ventilation and 64% died. In adjusted analyses, TMP/SMX was found to reduce the risk of PJP by 89%.

At face value, these findings would strongly support a role for PJP prophylaxis among patients with rheumatic diseases receiving rituximab, but subgroup analyses provide important nuance. Over half of the observed PJP cases occurred among patients with AAV, which conferred a four-fold increase in the risk of PJP. This is similar to prior investigations that suggested an elevated risk with AAV and would support the ACR/VF recommendations. The strongest risk factor in this study, however, was high dose glucocorticoid use, defined as over 30mg/day of prednisone for over 4 weeks. In adjusted analyses, high dose glucocorticoid use was associated with an 11–16 fold increase in PJP risk, and only one PJP case occurred among patients not-receiving high dose glucocorticoids.

The investigators also evaluated TMP/SMX-related adverse events and observed a 23% higher risk of adverse events in the prophylaxis group. However, most adverse events were of mild-to-moderate severity, including leukopenia, anemia, thrombocytopenia, and transaminitis. There were no instances of Stevens-Johnson syndrome. In comparison to the 64% mortality rate associated with PJP infection, there were no deaths from adverse events and all severe adverse drug reactions improved shortly after discontinuation of TMP/SMX. Further, only 12 of the over 300 adverse events in the prophylaxis group were causally attributed to TMP/SMX; the assessment of causality based on chart review is an important strength of this study.

To weigh the risks and benefits of prophylaxis, the investigators compared the number needed to harm (NNH) (i.e., cause a severe adverse event) to the number needed to treat (NNT) to prevent one PJP infection. Numerically, this comparison favored no prophylaxis, because 146 would need to be treated to prevent one infection as opposed to the 86 that would need to be treated to cause one severe adverse event. Presenting the balance of risks and benefits using NNT and NNH is appealing and can provide a simple tool for communicating risk to patients. However, comparing NNT to NNH suggests an equivalency between the benefits and harms that may not exist and was not observed in this study. None of the adverse events were fatal and all improved with discontinuation of prophylaxis, which contrasts sharply with the benefit (reduced mortality) of prophylaxis. If anything, this comparison favors PJP prophylaxis, as most clinicians would recommend a therapy that reduces mortality at the cost of a marginally higher rate of non-fatal adverse events.

The aforementioned risk stratification based on AAV status and high-dose glucocorticoid exposure is also an important moderating factor. Those who received high dose glucocorticoids had a low NNT (20), while those who did not had a very high NNT (250). Similar discrepancies of a lesser magnitude were observed for age (NNT 137 vs. 41 for younger vs older than 60), AAV (NNT 69 other diseases vs 26 having AAV), and interstitial lung disease (NNT 134 vs 36 for having ILD). These findings suggest a more nuanced conversation with patients is needed, where the risks are explained alongside the potential benefits and the individual patient’s risk factors are considered. Such conversations can be illuminating when they emphasize patient values. One patient may accept a risk of a mild rash or leukopenia if it could prevent a potentially life-threatening infection, for instance, while another may prefer the small risk of severe infection over more common adverse events and a greater pill burden.

An important unacknowledged limitation of this study was the incidence of PJP as compared to similar studies in other settings. A single center cohort study by Schmajuk and colleagues, for instance, examined a similarly heterogenous population of patients with rheumatic diseases over a similar number of patients years and observed zero cases of PJP, though only 40 (13%) received rituximab (12). A European study of 192 patients with AAV who received rituximab only observed a single case of PJP (5). In a single center cohort study of 997 patients with lupus, four (0.4%) cases of PJP were observed (13), two of which were in patients with comorbid HIV. Whether these observed differences in PJP incidence are due to geographical differences in PJP exposure or glucocorticoid prescribing, study design, or other unidentified differences is unknown. However, the relatively high incidence of PJP observed by Park and colleagues appears to be an outlier when compared to other contemporary observational studies. Given the already-high NNT observed in several subgroups, a lower incidence of PJP, even among patients with AAV, may alter the risk-benefit calculus for clinicians and patients.

Important unanswered questions remain. First, the role of prophylaxis is unclear for patients who remain on rituximab as maintenance therapy after completing a steroid course. In AAV, the ACR/VF guidelines imply that prophylaxis should be continued, and the observation of incident PJP after discontinuation of prophylaxis in this study led the authors to propose a similar approach. However, this study observed low rates of PJP among patients who did not receive high dose glucocorticoids, suggesting that the long-term rate of PJP after glucocorticoid discontinuation may be low. Importantly, the impact of continuous B cell depletion, often for years, also remains unknown. Second, practice patterns have evolved and patients with AAV who received the PEXIVAS-reduced steroid taper (14), LoVas-low steroid taper (15), or avacopan (16) instead of a usual steroid course would not have qualified as receiving “high dose” glucocorticoids. The role of PJP prophylaxis will need to be reconsidered as practice patterns evolve. Finally, the role of PJP prophylaxis in patients treated with rituximab for conditions other than AAV is less certain given the relatively low number of such patients in this study and the lower incidence observed among them.

This and other studies suggest an important role for PJP prophylaxis among patients with AAV who are receiving induction therapy with rituximab and high dose glucocorticoids. Those with strong risk factors for PJP, which include greater than 30mg of prednisone for over 4 weeks, age over 60, or structural lung disease, may be most likely to benefit. Whether the risks and benefits of PJP prophylaxis for similar subgroups will be observed in other conditions (e.g., giant cell arteritis, lupus) commonly treated with high-dose glucocorticoids, with and without other immunomodulators, requires additional studies. Though not studied here, the ancillary benefits of TMP/SMX in preventing other infections, as demonstrated recently in AAV, may also inform decision making (4, 5). In future studies, alternative approaches to communicating nuanced findings like these when engaging in shared decision making need to be tested to inform implementation of these findings.