This study reports a single-centre experience using aprotinin in a high-risk patient cohort undergoing iCABG surgery. The propensity score-matched analysis of 1026 patients, including 51 patients treated with aprotinin, revealed noteworthy findings. Propensity-score matching was not able to eliminate all baseline risk and comorbidity differences between the two cohorts. Despite significantly greater preoperative risks in the aprotinin group, such as increased rates of heart failure, poorer left ventricular ejection fraction, and higher EuroSCORE, there were no statistically significant differences in hospital deaths compared to the control group. Nevertheless, there was a 5.8% difference in the mortality rate between the aprotinin-treated patients and the control patients, which could indicate an element of clinical importance. Arguably, the sample size was too small to reliably determine the statistical significance of the differences.

A secondary analysis of the propensity score-matched data was also conducted to investigate the cause of mortality. The average EuroSCORE was 13 (range: 11–15) among the patients who died, which was significantly greater than that of the average cohort population. Furthermore, both instances of postoperative stroke and 2 out of 3 cases requiring new dialysis in the aprotinin cohort were among the patients who died before discharge. Perhaps even more significantly, surgery for 80% of patients who died in hospital were classified as emergency procedures.

The secondary outcomes of new renal replacement therapy and postoperative stroke were not significantly different, although they were more comparable between the two groups, and any difference could be due to worse preoperative conditions in the aprotinin group. Furthermore, concerns about poor renal outcomes and postoperative strokes may be partly linked to the nature and risk of surgery rather than exclusively the use of aprotinin. A large cohort study conducted by Furnary et al. suggested that poor renal outcomes were related to increased transfusions in high-risk patients, an established cause of renal dysfunction, and the absence of aprotinin use [13]. Furthermore, even in the BART study, aprotinin did not significantly affect the incidence of renal failure [9]. Although the aprotinin cohort had slightly greater rates of transfusion, this may be closely linked to the fact that most of these patients were urgent and did not stop DAPT or had platelet dysfunction prior to surgery.

The study also evaluated postoperative recovery periods between the two groups. These differences approached statistical significance, suggesting a possible trend toward longer recovery periods in the aprotinin group (p = 0.11 for ICU stay and p = 0.13 for hospital stay). However, since the baseline characteristics of the aprotinin cohort were worse than those of the control group, prolonged ICU and hospital stay may have been expected.

The reintroduction of aprotinin was associated with several conditions, including a limited licence in iCABG and a high risk of blood loss. Many studies have shown that aprotinin is safe for use in high-risk surgery. A large meta-analysis of more than 30,000 patients by Mehbohm et al. [14] showed that early mortality in high-risk patients did not differ between aprotinin and lysine analogue use. Other retrospective studies have also shown decreased mortality and reduced blood loss associated with aprotinin use in high-risk and open-heart procedures [15, 16]. Walkden et al. have also shown that since the withdrawal of aprotinin, mortality in high-risk cardiac surgery patients has significantly increased [17]. Furthermore, a large observational study conducted around the same time as the BART study showed that aprotinin did not adversely affect short- or medium-term survival [18]. These findings were further supported by a large-scale meta-analysis by Howell et al., which showed no increase in mortality with aprotinin compared to other antifibrinolytics [19]. According to Health Canada and the EMA, the therapeutic advantage of aprotinin may outweigh its risks in many cases and does not increase mortality [10, 11].

There are certain limitations to the studies since 2006, including the BART trial, which has been called into question regarding issues in the analysis, methods, and participant selection. Most of the studies were observational and included patients with a very high preoperative risk when aprotinin was used, increasing the bias against aprotinin when poor outcomes may have been the natural course due to the patients’ condition [20]. The BART trial also excluded iCABG patients and patients who underwent lower risk procedures. McMullan et al. mentioned unexplained exclusions of patients, which may have had statistically significant impacts on the results of the BART study [20].

Recently, the results from the analysis of the Nordic Aprotinin Patient Registry (NAPaR) have been discussed in the European Journal of Anaesthesia. The authors describe comparable outcomes between iCABG patients in the NAPaR cohort and those from the last 10 years. Notably, the baseline characteristics of the iCABG patients in the registry were more comparable to those of the patients in the control cohort of this study. Drawing from the conclusions of this study, NAPaR provides more concrete evidence regarding the lack of a causal link between aprotinin use and mortality as well as secondary outcomes [12]. However, the study did not use any matching or adjustment for baseline differences in patients before comparing outcomes. The addition of data to the NaPaR registry is key to improving the availability of evidence that allows for comprehensive evaluation of aprotinin in the future.

It is important to recognise that this is one of very few studies that has discussed the outcomes of aprotinin use since its reintroduction. The key difference in methodology is the use of logistic regression powered propensity score matching and the limitation of patients to iCABG surgery. This allows more relevant comparisons between the aprotinin and control groups in this cohort. Our results show statistically insignificant differences in the mortality and all other major postoperative complications unlike much of the evidence that was cited prior to aprotinin’s suspension. Most of these studies either showed increase in mortality or increased postoperative complications such as renal dysfunction or stroke [5,6,7,8,9].

The greatest limitation of the study is the small patient cohort with no medium-to-long-term data, which makes it difficult to assess mortality outcomes. Secondly, since this study was designed retrospectively after data collection, the paucity of reliable long-term data for most patients made it difficult to comment accurately on longer-term outcomes or use in survival analysis. Finally, although multivariable logistic regression-powered propensity matching was utilised to mitigate differences between the aprotinin-treated patients and control patients, the quality of the matching may not be ideal, as shown by the persisting risk difference in the matched cohorts. Unfortunately, the current practice of reserving aprotinin for use in high-risk patients makes mitigating this bias difficult [21].

The landscape of cardiac surgery is constantly changing. Even as early as 2005, research showed trends toward increasing rates of iCABG in high-risk patient groups. Patients who underwent surgery tended to be older and had more comorbidities [22]. This trend, as described by Dimeling G et al., has continued to increase in recent years [23]. More patients than ever are undergoing urgent surgical revascularisation and using dual antiplatelet therapies, leading to a higher risk of bleeding during surgery [24]. Moreover, surgical techniques and blood conservation strategies have made major leaps in the time since aprotinin was first suspended [25, 26]. This means that surgeons need reliable and contemporary evidence that supports the use of safe and effective antifibrinolytic therapy in high-risk surgery.

Further randomised trials with robust methods or a multicentre trial through analysis of a registry may allow for better mitigation of the baseline characteristics between patients. Future studies should also employ a power calculation to determine the minimum effect size. Future work should compare standardised, comparable patients who received aprotinin to those who did not, ideally through a large randomised controlled trial.