The study, led by Ju Hee Kim and Jungbin Song with colleagues at CHA University, Kyung Hee University, and Korea University Guro Hospital, examined three herbal formulations: Bojungikgitang, Ijintang, and Cheongsanggyeontongtang. All three have long histories of symptomatic use in East Asian traditional medicine for cardiovascular and cerebrovascular conditions—precisely the patient populations most likely to be taking anticoagulants. The research team designed a fixed-sequence, two-period clinical trial in which healthy volunteers received a single 60-milligram dose of edoxaban alone, and then received the same dose after repeated dosing with one of the three herbal formulations. The design allowed each participant to serve in part as their own control, isolating the effect of the herbal co-administration from natural interindividual variability in drug handling.

Edoxaban is a direct factor Xa inhibitor, a member of the direct oral anticoagulant family that has largely displaced warfarin in many markets because it requires no routine monitoring and has fewer dietary interactions. Its disposition in the body, however, depends heavily on the efflux transporter P-glycoprotein and, to a lesser extent, on cytochrome P450 3A4 (CYP3A4) metabolism, which produces an active metabolite known as M4. This pharmacological profile makes edoxaban particularly vulnerable to interactions with drugs—and potentially herbal phytochemicals—that inhibit or induce P-glycoprotein or CYP3A4. Strong P-glycoprotein inhibitors such as quinidine and cyclosporine are already known to raise edoxaban levels, while inducers such as rifampin can slash them, and the edoxaban label reflects those established interactions.

To measure what the herbal formulations did to edoxaban, the researchers quantified plasma concentrations of both the parent drug and its active M4 metabolite using liquid chromatography coupled with tandem mass spectrometry, a gold-standard bioanalytical technique aligned with international ICH M10 validation guidance. Pharmacokinetic parameters—the area under the concentration-time curve, peak concentrations, and clearance measures—were compared between treatment periods using linear mixed-effects models, a statistical framework well suited to repeated-measures crossover designs. In parallel, the team assessed pharmacodynamic effects using prothrombin time and activated partial thromboplastin time, the two coagulation assays most commonly used to gauge anticoagulant intensity at the bedside. Finally, in laboratory experiments, the formulations were tested for their capacity to inhibit CYP3A4 activity in vitro.

The pharmacokinetic results were striking in their divergence. Bojungikgitang approximately doubled systemic exposure to edoxaban, effectively reproducing the magnitude of interaction seen with known strong P-glycoprotein inhibitors. At the other extreme, Cheongsanggyeontongtang reduced edoxaban exposure by roughly 40 percent—a reduction in the range that regulators associate with inducers of the drug’s clearance pathways and one that could, in theory, erode the anticoagulant’s protective effect against stroke. Ijintang, by contrast, exerted only minimal effects on the drug’s pharmacokinetic profile. The active metabolite M4 followed the same directional pattern as the parent compound in each case, indicating that the herbal formulations were shifting edoxaban’s overall systemic disposition rather than selectively rerouting its metabolism.

Perhaps the most provocative—and most cautionary—finding came from the pharmacodynamic measurements. Despite the substantial pharmacokinetic shifts, the researchers detected no significant changes in prothrombin time or activated partial thromboplastin time in any of the herbal co-administration groups. In other words, a patient’s exposure to edoxaban could double, or fall by 40 percent, while the standard laboratory coagulation tests remained essentially silent. This dissociation matters because clinicians sometimes lean on coagulation assays when concerns arise about anticoagulant intensity, and the results suggest those assays are poor sentinels for herb-driven changes in edoxaban levels. The therapeutic window of a drug, not the laboratory value alone, ultimately determines whether exposure changes translate into bleeding or thrombotic risk.

The mechanistic work offered clues about how the divergent effects arise. In vitro testing showed negligible inhibition of CYP3A4 by Bojungikgitang and Ijintang and only weak inhibition by Cheongsanggyeontongtang, indicating that direct CYP3A4 blockade is unlikely to explain the doubling of edoxaban exposure seen with Bojungikgitang. The more plausible culprit is interference with P-glycoprotein, the efflux pump that governs edoxaban’s intestinal absorption and biliary excretion. Many phytochemicals common in traditional formulas—including compounds found in ginseng, licorice, and astragalus, herbs represented in some of these formulations—have been shown in laboratory studies to modulate P-glycoprotein and CYP enzymes. The precise constituent responsible for the Bojungikgitang effect, however, remains to be pinned down, and the authors’ data do not permit attribution to a single botanical ingredient.

The clinical context amplifies the significance of the findings. Surveys of Korean medical hospital practice have documented frequent combination therapy between herbal extracts and antiplatelet or anticoagulant drugs in ischemic stroke patients, and Korean medicine clinical practice guidelines for stroke, developed under the National Institute for Korean Medicine Development, formalize herbal care pathways for the same patients who often need anticoagulation. Case reports from around the world have described increased bleeding in patients combining oral anticoagulants with herbal products ranging from saffron to Salvia miltiorrhiza, and St John’s wort has been shown to lower rivaroxaban exposure in controlled human studies. Meanwhile, cohort data have suggested that some concurrent Chinese herbal medicine use may even be associated with fewer major bleeding events, underscoring that the herb-drug interaction landscape is neither uniformly dangerous nor uniformly benign—it is formulation-specific, and largely unmapped.

That is precisely the gap this trial helps to fill. By testing three named, commercially supplied formulations under controlled conditions in healthy volunteers, and by measuring both drug levels and clotting outcomes with validated methods, the Korean team has converted a vague anxiety—herbs might interact with blood thinners—into concrete, quantifiable risk statements: Bojungikgitang approximately doubles edoxaban exposure, Cheongsanggyeontongtang cuts it by about 40 percent, Ijintang is relatively benign, and conventional coagulation tests will not warn you about any of it. The authors conclude that clinical consideration is warranted when edoxaban is combined with Bojungikgitang or Cheongsanggyeontongtang, and the magnitude of the observed changes supports that caution, situating both interactions on par with labeled drug-drug interactions of edoxaban.

Important caveats temper the translation to patient care. The study enrolled healthy volunteers, not the older, multimorbid patients with atrial fibrillation or venous thromboembolism who actually take edoxaban, and it could not assess hard clinical endpoints such as bleeding or stroke. Fixed-sequence designs and modest sample sizes, typical of exploratory interaction studies, also limit generalizability, and the duration of herbal pre-dosing may not capture longer-term induction or adaptation effects. Even so, the work delivers a clear, actionable message for prescribers, pharmacists, and practitioners of traditional medicine alike: herbal formulations are not pharmacologically inert companions to modern anticoagulants. Until larger outcome studies are done, patients taking edoxaban who also use Bojungikgitang or Cheongsanggyeontongtang deserve explicit attention, honest conversations about herbal supplement use, and a healthy skepticism toward coagulation tests that may not reveal what the body is actually doing.

Subject of Research: Herb-drug interactions between traditional Korean herbal formulations and the anticoagulant edoxaban

Article Title: Effects of Bojungikgitang, Ijintang, and Cheongsanggyeontongtang on the pharmacokinetics and pharmacodynamics of edoxaban

Article References: Kim, J. H., Song, J., Kim, M., Kim, H., Cho, H.-Y., & Chung, H. (2026). Effects of Bojungikgitang, Ijintang, and Cheongsanggyeontongtang on the pharmacokinetics and pharmacodynamics of edoxaban. Journal of Pharmaceutical Investigation. https://doi.org/10.1007/s40005-026-00828-z

Image Credits: AI Generated

DOI: 10.1007/s40005-026-00828-z

Keywords: edoxaban, herb-drug interaction, Bojungikgitang, Ijintang, Cheongsanggyeontongtang, pharmacokinetics, pharmacodynamics, direct oral anticoagulants, P-glycoprotein, CYP3A4, traditional Korean medicine, clinical trial

What happened next transformed this routine infection into a diagnostic puzzle that physicians rarely encounter. By hospital day 11, imaging showed that the right upper lobe infiltrate had resolved, but new bilateral pleural effusions had appeared in its place. Thoracentesis was performed on both sides of the chest, and both effusions were grossly bloody and exudative. Yet beneath that identical bloody appearance lay a startling discordance: the right-sided fluid was dominated by neutrophils, at 85 percent, while the left-sided fluid was dominated by lymphocytes, at 90 percent. Cytology was negative for malignancy in both samples, and cultures grew no organisms. A single systemic process, such as heart failure or carcinomatosis, could not plausibly produce two fluids with such different cellular signatures. Instead, the findings pointed toward two entirely independent diseases developing simultaneously in the same chest.

The right effusion was straightforward to explain. Parapneumonic effusions are a well-recognised complication of bacterial pneumonia, particularly pneumococcal infection, in which parenchymal inflammation extends to the pleural space and generates a protein-rich, neutrophil-dominant exudate driven by acute inflammation. The antibiotic therapy was working against the pneumonia itself, but the severe initial inflammatory burden likely led to a delayed manifestation of pleurisy, with effusions emerging even as the lung infiltrate retreated. That mechanism accounted for the right hemithorax. The left effusion, with its lymphocyte predominance and later behaviour, was a far more elusive entity that would take months to reveal its full character.

The bloody appearance of both effusions added another layer of complexity. The patient’s antithrombotic regimen included edoxaban, a direct oral anticoagulant, along with aspirin, prescribed after coronary artery bypass grafting approximately 12 months earlier for acute coronary syndrome. Direct oral anticoagulants have been reported to induce spontaneous pleural bleeding, and the clinicians hypothesised that the combination of anticoagulant and antiplatelet therapy exacerbated microvascular oozing within inflamed pleura. Notably, however, the red blood cell counts in the effusions were relatively modest, yielding calculated pleural fluid haematocrits of only 0.18 percent on the right and 0.44 percent on the left, far below the diagnostic threshold for haemothorax. In other words, the bleeding tendency likely painted the exudates red without fundamentally changing their underlying inflammatory nature.

Management reflected the uncertainty. Given the bloody effusions, the team discontinued only the edoxaban and administered haemostatic agents, while continuing aspirin after consultation with the cardiovascular surgery team. Follow-up chest CT on May 21 showed partial resolution of the effusions, and with fever and hypoxaemia resolved, the patient was discharged on May 23. He remained stable as an outpatient for three months. Then, in late August, the left effusion worsened, and a concurrent chest CT revealed a small pericardial effusion that was too small for pericardiocentesis. A repeat left thoracentesis again yielded bloody, lymphocyte-dominant fluid, and C-reactive protein, which had normalised after discharge, had crept back above the institutional upper limit of normal. Echocardiography demonstrated preserved left ventricular function, arguing against heart failure as the cause.

This late recurrence allowed the clinicians to apply formal diagnostic criteria. The 2015 European Society of Cardiology guidelines for pericardial diseases require at least two of five criteria for a diagnosis of post-cardiac injury syndrome, the category that encompasses Dressler syndrome. The patient fulfilled two: pericardial effusion on CT and pleural effusion with elevated inflammatory markers. Fever, pericarditic or pleuritic chest pain, and audible rubs were all absent, making the presentation distinctly indolent. Late-onset Dressler syndrome was suspected, and the team initiated diagnostic therapy with ibuprofen at 600 mg per day, the maximum dose covered by Japanese medical insurance. Follow-up imaging confirmed significant improvement of both the left pleural and pericardial effusions, and no recurrence occurred during 10 months of follow-up after the anti-inflammatory treatment began.

Dressler syndrome is an immune-mediated pleuropericarditis that classically appears within weeks to a few months of cardiac injury, most often after cardiac surgery such as coronary artery bypass grafting. The prevailing pathophysiological model holds that cardiac injury releases sequestered cardiac antigens, triggering the formation of anti-heart autoantibodies. The time required for this immunological sensitisation inherently creates a latent period between injury and clinical onset. The reported profile of post-cardiac injury syndrome effusions fits this patient’s left-sided fluid almost exactly: such effusions are unilateral in about two thirds of cases and left-sided in the large majority, with no Dressler syndrome patients in one published series showing right-sided unilateral effusions; nearly all are exudates, and roughly three quarters are lymphocyte-predominant.

The timing, however, was extraordinary. Onset occurred approximately 12 months after bypass surgery, exceeding not only the usual latent window but also the nine-month interval described in one of the longest previously reported cases. The authors propose a two-hit hypothesis: the patient developed long-standing latent immune sensitisation following his cardiac surgery, and the intense systemic inflammatory response generated by severe pneumococcal pneumonia acted as a necessary catalyst, reactivating the dormant autoreactive process and culminating in the delayed manifestation of Dressler syndrome. In this framing, the pneumonia did not merely coexist with the immune syndrome; it may have ignited it. The two mechanisms are not mutually exclusive, and an immune-mediated predisposition related to prior cardiac injury may have been unmasked by systemic inflammation from the contralateral infection.

The authors are careful to acknowledge residual diagnostic uncertainty, since a response to non-steroidal anti-inflammatory drugs is common in pleuropericarditis of virtually any cause. A reactive post-infectious serositis following the pneumonia remains a plausible alternative, as autoimmune pleuropericarditis has been reported after community-acquired pneumonia, including Chlamydia pneumoniae-associated cases. Furthermore, no pericardial effusion was present at the time of the initial thoracenteses, and the left effusion was attributed to a post-cardiac injury process only retrospectively, once the August recurrence fulfilled the formal criteria. The lymphocyte-predominant profile and prompt ibuprofen response support an immune-mediated process, but they do not definitively tie it to the prior surgery rather than the preceding infection. Late-onset Dressler syndrome is therefore presented as the most likely clinical diagnosis rather than a certain one.

The broader lesson reaches well beyond this single patient. Clinicians frequently assume that bilateral pleural effusions stem from a single systemic aetiology such as heart failure, renal failure or malignancy, and this assumption can obscure independent localised pathologies that demand separate therapeutic strategies. In this case, the two sides of the chest required targeted antibiotic therapy on the right and anti-inflammatory treatment on the left, and neither intervention alone would have addressed both conditions. The definitive clue was not the grossly bloody appearance, which was identical on both sides and heavily influenced by antithrombotic medications, but the rare and striking discordance of pleural fluid cellular profiles between hemithoraces. For patients with bilateral effusions whose clinical or radiological presentation is atypical, the case argues strongly for bilateral thoracentesis with detailed cellular analysis, a practice this team believes should be adopted far more widely. To their knowledge, a PubMed search identified no previous report of parapneumonic effusion and Dressler syndrome presenting concurrently, and the combination is classically described under the label of Contarini syndrome, in which each side of the chest harbours a different cause.

Subject of Research: A rare case of bilateral bloody pleural effusions arising from two independent causes, concurrent parapneumonic effusion and suspected late-onset Dressler syndrome, distinguished by discordant pleural fluid cellular profiles.

Article Title: Bilateral Bloody Pleural Effusions With Discordant Cellular Patterns: A Case of Concurrent Parapneumonic Effusion and Dressler Syndrome

Article References: Narumi, Y., Hoshino, S., Yamakawa, M., & Amano, Y. (2026). Bilateral Bloody Pleural Effusions With Discordant Cellular Patterns: A Case of Concurrent Parapneumonic Effusion and Dressler Syndrome. Respirology Case Reports, 14(9), Article e70764. https://doi.org/10.1002/rcr2.70764

Image Credits: AI Generated

DOI: 10.1002/rcr2.70764

Keywords: pleural effusion, Dressler syndrome, post-cardiac injury syndrome, parapneumonic effusion, Streptococcus pneumoniae, thoracentesis, coronary artery bypass grafting, edoxaban, lymphocyte-dominant exudate, Contarini syndrome, pericardial effusion, ibuprofen