aHUS treatment protocols are changing globally in response to new clinical evidence. This mixed-methods study, comprising of qualitative interviews with international aHUS treatment experts and a quantitative decision tree analysis, sought to inform optimal treatment duration guidelines for adult patients with aHUS. The qualitative study is the first to systematically explore the complexity of factors influencing treatment discontinuation decisions in adult patients with aHUS. In terms of generalizability of these opinions and ability to incorporate this information in general practice, it is recognized that all interviewees were part of large university-based hematology or nephrology practices. As such, they had the capacity to do complement mutation analyses either in-house or as send-out tests. They also had access, usually within 12–24 h, to eculizumab and ravulizumab, enabling rapid drug re-initiation once discontinued. As aHUS is a rare and complex disorder, clinicians treating such patients or responsible for their follow-up, must either have similar access, or refer patients to facilities with such expertise and access. In addition, we limited participants to those nephrologists and hematologists treating adult aHUS patients, to remove age as a complicating factor in these discussions. For example, as described in a recent review [20], a large French aHUS registry study found that the number of relapses following resolution of renal dysfunction and a median treatment duration for eculizumab of 17.5 months was higher among adults than children, thought to be a consequence of divergent types of complement-related mutations in the two populations, but a multinational observational study reported off-treatment TMA rates lower, not higher, in those > 18 years of age. A parallel investigation of treatment duration decisions among pediatric aHUS experts would be valuable.

Our results suggest that, although the majority of interviewees were in favour of treatment discontinuation (citing the high cost, infusion burden, and potential side-effects of lifelong treatment as key reasons), a prior negative experience of discontinuation seemed to make others more reluctant to stop. Deciding which patients required lifelong treatment and which not, involved the consideration of a wide range of factors, including not only clinical (e.g., presence of mutations conferring a high risk of relapse), but also patient- and system-related factors (e.g., patient adherence to monitoring, availability of funding for re-treatment). Adherence to at-home monitoring for proteinuria, hematuria, and increased blood pressure, enabling rapid re-introduction of drug (within 7 days but ideally within 48 h) to preclude irreversible tissue injury, was stressed by all participating experts. For the patient this meant “having confidence” that they are being monitored effectively, while the physicians stressed their own assessment of an individual’s reliability to complete monitoring protocols. In terms of facilitating such adherence, a majority of experts noted the need for funding from health ministries to enable its oversight. While specifics of possible monitoring protocols were not provided, nor were they requested during these interviews, given that most aHUS relapses post-treatment occur within the first three months [20], many aHUS treating physicians rely on frequent (twice weekly for the first month, then weekly for the next two months) at-home testing documented by appropriately timed nurse or physician assistant telephone checks, accompanied by office visits at varying intervals for laboratory testing.

Our finding that a prior negative experience with discontinuing treatment influenced treatment duration was surprising, but perhaps should not have been. Non-clinical influences on clinical decision making are well-recognized impediments to evidence-based practice, particularly in the context of rare conditions when what amounts to single case reports – anecdotal testimonies from physician colleagues or the physician’s own practice – assume an outsized influence on a medical decision [29]. Our study argues for an international consensus conference to explore issues surrounding aHUS treatment duration.

The simple treatment duration decision tree presented in Fig. 1 estimated the potential reduction in time off treatment for an average adult aHUS patient, as well as the potential adverse risks of events for both treatment (serious infection such as meningitis) and discontinuing treatment (increased risk of TMA and resulting renal damage). It is clear that, for patients with aHUS, the risk of TMA, when off treatment, is increased relative to being on treatment [28]. Yet, with close attention to self-monitoring, re-initiation of anti-complement therapy can resolve the TMA with very little risk of long-term kidney damage. While this encourages the consideration of treatment discontinuation, other long-term studies note the poorer kidney outcomes for those discontinuing anti-complement therapy compared to those remaining on treatment [30].

The close monitoring requirements of treatment discontinuation relies primarily on the self-administration of urine dipstick tests, along with physician visits for blood workup in the event of infection episodes [20]. Since these tests detect kidney damage after it has occurred, they are less than ideal and require prompt re-initiation of treatment to reverse kidney damage. To date, much of the concern about high-risk patients has been related to the rate of TMA/relapse, but an alternative definition of a high-risk patient might consider those whose kidney function is already compromised (e.g., stage IV CKD patients) and for whom further renal injury could result in the need for dialysis or transplantation.

Although a very serious condition, the incidence of meningococcal meningitis with anti-C5 therapy (including for paroxysmal nocturnal hemoglobinuria) is low, about 0.6 event/100 patient years [27]. Yet, meningococcal prophylaxis prior to initiating anti-C5 therapy, and continued vigilance for symptoms of meningococcal infection during treatment and for a period post treatment discontinuation, is critical. Specifically, as mandated by the U.S. REMS (Risk Evaluation and Mitigation Strategy) program, all individuals must be vaccinated against Neisseria meningitidis serotypes A, C,W, Y and B 14 days prior to initiation of eculizumab or ravulizumab. If this is not feasible, then 14 days of prophylactic antibiotics (e.g., ciprofloxacin, phenoxymethylpenicillin, rifampicin) should be started with vaccination on the day of anti-C5 initiation. For individuals who are immunocompromised, or in areas with a high incidence of meningococcal serotype B, it may be prudent to continue prophylactic antibiotics for the duration of anti-C5 therapy. Vaccine boosters should be administered as per U.S. ACIP (Advisory Committee on Immunization Practices) guidelines. Furthermore, in the context of treatment discontinuation decisions, given the known duration of action of eculizumab vs. ravulizumab, vaccine protection and/or prophylactic antibiotic use should be continued for 3 months or 8 months post-eculizumab or ravuilizumab discontinuation, respectively. Risk of infection is, however, outweighed by the benefits of reduced TMA events.

The principal benefits of anti-C5 discontinuation is patient convenience in terms of no longer requiring infusion and the financial benefits (to the system and potentially the patient in the form of co-pays) of discontinuing what is a very financially burdensome therapy. The yearly cost of eculizumab in the US is circa $600,000 including administration, though this can be reduced by approximately 1/3 with a switch to ravulizumab, due largely to the reduced dosing schedule from 2-weekly to 2-monthly [31]. Nevertheless, this circa $400,000 cost of ravulizumab is still a huge burden and the model here shows this could be reduced by a further 79% (59% for high-risk patients) to circa $80,000 ($160,000) by discontinuing therapy in those that remain in remission. In this context, it should be noted that shortly after completion of these interviews, two interchangeable biosimilars to eculizumab were approved by the EMA, Bikemv (eculizumab-aeeb) (Amgen) and Epysqli (Samsung Bioepic). The former was also recently approved by the FDA and will be available in the U.S. in March 2025. The magnitude of expense linked to eculizumab use is thus expected to decline worldwide—one rationale for its FDA approval, to “support a competitive marketplace” [32]. In terms of patient quality of life, they still must be administered bi-weekly, as opposed to ravulizumab, which is infused bi-monthly.

Given the PI’s knowledge of the use of anti-complement therapies by all 25 experts initially approached, either from his personal interactions with such experts or review of their publications, we are confident that the 10 interviewed experts represented groups without a bias against long-term C5 blockade, regardless of what parameters to determine treatment duration were utilized. However, it was clear that one dissenting opinion, held by at least two of the 15 declining to be interviewed, is missing from this qualitative study: the concept that presence of a renal allograft should not be a strong criterion for life-long use of anti-complement treatment. This is based on their national cost-effectiveness considerations, noting that while failure to continue such therapy appears to result in an inferior survival rate of such allografts, patient survival was not threatened. It is a minority opinion internationally which, as noted in a recent review [20], contradicts a consensus statement from an international group of experts that “transplant patients, especially those who have lost previous allografts, are not good candidates for treatment cessation”.

The main limitation of the decision tree approach to treatment duration was to limit the analysis to the first-year post discontinuation when anti-complement therapy is potentially lifelong. The event rates and financial consequences identified in this first year could form the basis of a longer-term extrapolation model, similar to that conducted by Wang and colleagues [31]. Yet, it should be clear that any such extrapolation model would simply project the identified annual events and savings into the future while the subjects remain alive. With potential savings of circa $320,000 per year for the average patient, any lifelong model would predict many millions of dollars-worth of financial savings over subject lifetimes, with a similar annual risk of adverse events.