Half-Dose Versus Full-Dose Alteplase Therapy in Pulmonary Embolism: Does Half Dose Really Lose?

To the Editor: In a recent issue of Critical Care Medicine, we with great interest read the article by Kiser et al (1) which compared the effectiveness and safety of half-dose (50 mg) versus full-dose (100 mg) alteplase for treatment of pulmonary embolism (PE). The use of thrombolytic therapy remains controversial in patients with intermediate-risk PE for potential risks of bleeding events. Kiser et al (1) reported an increased treatment escalation with similar hospital mortality and rates of major bleeding in half dose compared with full dose through propensity score matching and detailed subgroup analyses, which seemed to inflict a forceful strike on half-dose alteplase therapy. However, several influencing factors remain, it is too early to blow the final whistle. First, increased treatment escalation, commonly seen in half dose, was driven by secondary thrombolysis and catheter thrombus fragmentation. As was mentioned in their article (1), it was unclear whether secondary thrombolysis and thrombosis fragmentation by catheter was preplanned or as a result of deterioration. A safer way of using alteplase for treatment of PE recommended in China is that 50 mg is pumped in the first hour to observe whether there is any adverse reaction, if not, another 50 mg is sequentially pumped in the second hour. Under the circumstances, the latter dose cannot be categorized into secondary thrombolysis, those patients actually should be grouped into preplanned full dose. Additionally, for those in danger who will likely die before thrombolysis can take effect, catheter thrombus fragmentation may be administered at the same time or following thrombolysis as a preplanned manner. Thus, the authors should exclude the aforementioned conditions which have a great impact on the results. Second, the risk stratification of PE is debatable. Kiser et al (1) differentiated high-risk and intermediate-risk PE only based upon vasopressor usage, which undoubtedly increased the number of high-risk patients among intermediate-risk PE subgroup, thus this grouping strategy might produce selection bias and theoretically enhance the mortality rate of intermediate-risk subgroup. Levels of brain natriuretic peptide and troponin, two good predictors of adverse outcome in patients with PE (2, 3), should be considered in risk stratification, whereas other data were not available. (3). “The primary finding of our article is that significant and potentially confounding association between predictor variables exists”: patients with vague symptoms were significantly more likely to experience antibiotic delays, and both were associated with mortality in univariate analysis. Therefore, we concluded that “the nature of presenting symptoms is an important component of sepsis clinical phenotyping and may be an important confounder in sepsis epidemiologic studies.” It is possible, under extreme cases of collinearity, also termed “multicollinearity,” that is, two predictors are functionally redundant, that a regression model becomes mathematically unstable. As can be seen in Table 1 in (2), the association between vague/explicit symptoms and time-to-antibiotics does not suggest that the variables provide mathematically redundant information. This was confirmed by calculating the correlation between covariates using different cutoffs for antibiotic delay. A companion report, published in the same issue of Critical Care Medicine by Rhee et al (4), analyzed 851 septic patients from seven different medical centers. The term “vague/explicit” symptoms as defined by our methods was an independent predictor of mortality. Statistically significant benefit of timely antibiotics was present only in those who presented with explicit symptoms (odds ratio [OR], 3.35; 95% CI, 1.08–10.4), and not for patients with vague symptoms (OR, 1.50; 95% CI, 0.71–3.17), demonstrating the importance of vague symptoms in the relationship between time-to-antibiotics and mortality. For our report, the timing of antibiotics was extracted from the nursing documented administration time. Regarding our definition of “explicit” and why shortness of breath was not included: as stated in the article (2), we developed that definition a priori based on clinical judgment, on the assumption that “shortness of breath” was a nonspecific symptom that could be caused by many noninfectious etiologies. We agree with Karakonstantis et al (1) that early recognition of sepsis remains imperative for sepsis treatment. Drs. Filbin's, Reisner's, and Heldt's institutions received funding from Nihon Kohden Corporation, Controlled Risk Insurance Company Risk Management Foundation, and Massachusetts General Hospital/Massachusetts Institute of Technology Grand Challenge Award. Dr. Filbin received grant funding from Rapid Pathogen Screening Incorporated and ThermoFisher Scientific. Dr. Reisner received funding from Covington and Burling, LLP, and Eli Lilly (stock holdings,<$5k).