Discussion
In this systematic review, we identified 315 infections in 290 patients with VAs undergoing treatment with sirolimus. Notably, serious infections, infection-related deaths, bacteremia, and PJP were infrequent. The vast majority of infections were respiratory infections, with 54% upper respiratory infections and 20% pneumonias. Overall, 25% of all patients reported in evaluable studies had an infection reported, and fewer than 10% of all reported patients had serious adverse events related to infections (CTCAE grade 3 or greater). Furthermore, it is unclear how many of these infections were attributable to sirolimus. For example, 54% of infections were upper respiratory viral illnesses. While sirolimus could theoretically increase the risk of developing these viral infections, such infections are also common in the general pediatric and adult population. Furthermore, cutaneous infections (or inflammation mimicking cutaneous infections) often occur on the skin overlying vascular malformations, and sirolimus is used to decrease the frequency of these flares.26 As such, the 20 reported cutaneous infections could be related to the underlying VA, rather than as a result of sirolimus’ immunosuppressive effects, and they might not actually represent infections.
All reported fatalities occurred in children 2 years of age or younger, and were most common in infants and related to pulmonary infection. These deaths generally occurred within 2 months of starting sirolimus. Furthermore, the majority of these deaths occurred in infants with KHE who also had KMP, a consumptive coagulopathy.4 While these findings could underrepresent the true frequency of infection-related fatalities due to insufficient reporting, they suggest that death is very infrequent, and infants with KMP or other complications are most at risk. Furthermore, they suggest that clinicians should maintain vigilance for respiratory infections and pneumonia in patients treated with sirolimus. These data also leave several questions unanswered. Future studies should address whether patients with KHE have a higher risk of infectious complications with sirolimus usage, whether concomitant steroid usage is a significant risk factor, and whether patients are at a higher risk of serious infectious complications soon after initiating sirolimus therapy. Furthermore, studies should evaluate whether KMP or other complications represent a selected patient population that is at increased risk of infection due to comorbidities.
Bacteremia was a very uncommon infectious complication. Of all 1181 patients reported in these studies, 0.2% of patients developed bacteremia. Furthermore, bacteremia accounted for 0.6% (4/315) of infections ever reported in this literature. Of these 4 cases, 2 patients had central lines that served as a likely nidus for infection. For the remaining 2 patients, there was no clear source of the infection reported in the studies. For patients without a central line, these data suggest that bacteremia is a rare occurrence and these patients do not routinely require blood cultures and antibiotics with febrile episodes. Patients with a central line, however, should continue to be evaluated for central line associated bacteremia with fevers, as is standard of care for all patients with an indwelling central catheter.
We only identified one patient who developed PJP while receiving sirolimus (0.08% of 1181 total patients reported in these studies), and this patient was an infant with KHE and KMP who was weaning off steroids. Steroids alone in infants can increase the risk of PJP, thus warranting PJP prophylaxis in this patient.27,28 As such, PJP is an extremely rare event for patients with VAs receiving sirolimus. Although many patients in these studies likely received PJP prophylaxis, our findings should encourage a re-evaluation of the risks and benefits of PJP prophylaxis. Given the documented PJP in an infant, it would be reasonable to encourage prophylaxis for infants with VAs receiving sirolimus, especially if they have complicating factors like steroid treatment, KMP, or other comorbidities. However, these data suggest that PJP prophylaxis might not be warranted for many patients with VAs receiving sirolimus without significant comorbidities or concomitant immunosuppression. Future multicenter studies should aim to assess the incidence and risk factors of PJP using retrospective cohort studies, informatics queries, or claims data to better guide prophylaxis decisions.
This study should be interpreted in light of limitations. First, we excluded abstracts to avoid incomplete data or duplication with subsequent publications. However, abstracts could have contributed additional evidence of infectious complications. Furthermore, the frequency of published reports of infections does not necessarily represent the clinical experience with infections in this patient population. Reporting bias could result from either increased reporting of infections (overestimation) or decreased reporting of infections (underestimation). Furthermore, several key characteristics of infections were omitted from included studies. As such, we are limited in our ability to generate hypotheses about which patient characteristics might place patients at higher risk of infectious complications.