July 2026 Research Update

This month added a clearer picture that some of the hardest practical decisions in PNH are still handled very differently across centers, especially around blood thinners and what to do during treatment changes. A survey across Italy showed major differences in when centers use preventive anticoagulation, while a multicenter switch study showed that changes between complement inhibitors can be made with careful continuity but are not always smooth. Together, these studies emphasize that the day-to-day "how" of PNH management can vary even in the complement-inhibitor era, and that certain transition periods deserve extra attention in research and practice. [1] [2]

In 24 Italian centers caring for 291 people with PNH, approaches to preventing clots before starting complement inhibitors ranged from routine use in about one third of centers to no use in about one third, with others deciding case by case. After a clot had already happened, centers reported using anticoagulation for everyone, and about 75% said they continued it indefinitely even if the person was also on complement inhibition. During breakthrough hemolysis (a flare of blood cell breakdown while on treatment), centers also differed on whether and how long to use anticoagulation, often tying duration to lab improvement. [1]

A separate international report focused on what happens when people switch from one complement inhibitor to another. Across 65 patients and 149 treatment changes, short-term hemolytic events within 14 days were uncommon for some switch types but were reported more often when switching from a proximal inhibitor to a terminal inhibitor (13 of 29 switches, 44%). No thrombosis or death was reported during switching in this dataset, but some people needed transfusions around these events, underscoring that switching can be a vulnerable moment even when serious outcomes are not seen. [2]

In a different setting entirely, a large electronic health record study compared people with septic shock who received a terminal C5 inhibitor (eculizumab or ravulizumab) with matched people who did not; some had PNH. They reported no difference in 30-day mortality between the two groups, but did see differences in rates of kidney and blood-related complications and thrombotic events. Because this study was not limited to PNH and focused on septic shock, it mainly adds context about outcomes in a critical illness scenario rather than everyday PNH care. [3]

For patients and caregivers, these studies help explain why different centers may handle clot prevention, anticoagulation duration, and treatment transitions differently in PNH. The survey and switch study are real-world reports, so they describe what happened in practice rather than proving which approach is best, but they do highlight specific situations that may carry short-term risks (like certain drug switch directions) and areas where standard approaches are still being worked out. The septic shock comparison is most relevant if someone with an indication for C5 inhibition becomes critically ill, and it does not by itself answer what would happen in stable PNH outside that setting. Overall, this months evidence mostly adds understanding of variability and transition risks, and it points to the need for clearer, consistent strategies in future research. [1] [2] [3]

This month strengthened the message that delays in recognizing PNH remain common, but that newer ways of searching for it may help, especially when symptoms do not look classic. An AI screening program in Poland found more confirmed cases per referral than conventional approaches, and case-based reports from Kenya and from people with lupus showed how easily PNH can hide behind other explanations like chronic anemia, bone marrow failure, or autoimmune disease. Together, these studies make the diagnostic picture feel more concrete: the challenge is often not the test itself, but knowing who to test and when. [4] [5] [6]

In Poland, an AI-based tool screened electronic health records from 14 health organizations, reviewing over 1.3 million patients to flag people at high risk for PNH. Among 119 people referred for confirmatory flow cytometry, 13 were diagnosed, for a positive predictive value of 10.92%, which the authors reported as higher than the conventional screening hit rate of 6.9%. The strongest predictors included Coombs-negative hemolysis (blood breakdown not explained by typical autoimmune hemolysis testing) and frequent healthcare visits, suggesting the tool may help identify people whose presentations are easy to mislabel. [4]

The Kenya case-series of seven young adults put a human face on diagnostic delay and access barriers. All had PNH confirmed by flow cytometry, most had very large PNH clones and needed transfusions, and the average diagnostic delay was about 5 years. Complement inhibitor therapy was not used long-term because of limited access, and care was largely supportive, showing how outcomes and options can differ sharply by setting even when the diagnosis is clear. [5]

A lupus-focused case report with a review of seven similar cases highlighted that even small PNH clones can matter when a person has ongoing low blood counts or Coombs-negative hemolysis. In the collected eight cases, most had hemolysis and some had thrombosis, even when clone sizes were small in certain blood cell lines. The index patient improved on lupus-directed therapy over more than 2 years while small PNH clones persisted, reinforcing that PNH and autoimmune disease can overlap and may require high-sensitivity testing to detect. [6]

For patients, the main practical message is that PNH can be hard to recognize, especially when symptoms resemble other causes of anemia or low blood counts, or when another condition like lupus is present. The AI screening work is a real-world diagnostic approach that may matter in the future if it can be implemented safely and broadly, but it is not yet a standard part of routine care everywhere. The case-based reports do not prove how often these situations happen, but they make it clearer that long delays, limited access to advanced therapy, and overlooked PNH clones are real experiences for some people. What remains uncertain is how well AI tools and broader high-sensitivity testing will perform across different healthcare systems and how often small clones lead to major problems in larger groups. [4] [5] [6]

This month included one study that did not test treatments in patients, but instead examined how health technology assessments handle uncertainty when treatments look similar. It found that many NICE cost-comparison evaluations do not use formal tools to interpret non-significant indirect comparisons, and it proposed a Bayesian plotting approach to make uncertainty easier to see. Using PNH as an example, it showed how crovalimab comparisons with eculizumab and ravulizumab could be expressed as probabilities of being non-inferior, rather than treated as simply "no difference." [7]

In the crovalimab case example (NICE TA1019), the indirect comparison estimates versus eculizumab and versus ravulizumab had wide uncertainty ranges that crossed no difference, yet the framework produced probabilities of non-inferiority of 95.87% and 86.27%, respectively. The paper also reported that only about 32% of evaluated NICE appraisals used formal methods to explore this kind of uncertainty, and about 75% did not use clear non-inferiority margins. The key point is about how evidence is interpreted for decisions, not about proving that one PNH drug works better for patients. [7]

For patients and caregivers, this study mainly explains why you may hear different-sounding conclusions from similar data, especially when a new medicine is compared indirectly rather than head-to-head. It adds a way to describe uncertainty in plain terms (for example, the probability that one option is not meaningfully worse), which could matter later in how coverage or value decisions are communicated. It does not provide new information about symptoms, safety, or quality of life for crovalimab, eculizumab, or ravulizumab in people living with PNH. What remains uncertain from this paper alone is how these interpretation tools would change real-world decisions across different countries and appraisal processes. [7]