The purpose of hematopoietic stem cell transplantation (HSCT) in the context of malignant diseases is to regenerate bone marrow function, often compromised by prior conditioning regimens, and/or to trigger a graft versus tumor effect1,2. Conditioning regimens may result in significant complications3, which tend to increase the patients’ drug intake and the potential for drug-drug interactions (DDIs). The most worrisome DDIs are those that have a negative effect on the patient. As such, these DDIs must be identified, prevented and resolved4,5.

Patients undergoing HSCT are usually treated with complex pharmacological regimens, which may lead to multiple DDIs, increasing the risk of adverse events or reducing therapeutic effectiveness. In fact, several authors have reported severe adverse events such as rhabdomyolysis6,7 and even life-threatening complications in the context of HSCT8. Other factors that may contribute to increasing the risk of DDIs include the number of drugs administered, the length of hospital stay and the type of procedure performed. Moreover, these patients present with associated comorbidities including renal and hepatic dysfunction, an impaired nutritional status and protein-binding displacement, which increases the risk of clinically significant DDIs9–11.

The main purpose of this article was to carry out a systematic literature review to identify studies on DDIs occurring in patients undergoing HSCT. Secondary goals included a description of the prevalence of such DDIs and the collection of data on specific DDIs.

A structured literature review was carried out using PubMed and Embase. The goal was to identify as many original articles as possible dealing with DDIs in patients undergoing HSCT following the PRISMA methodology. The analysis included clinical trials, observational studies, case reports or original case series, and letters to the editor. To be included, studies had to report results directly associated with the purpose of the review, the had to have been written in either English or Spanish, and their publication date had to be comprised between 1 January 2000 and 27 November 2020. Publications not related to DDIs and those related to DDIs but not to HSCT, as well as articles published only as oral papers for submission to a conference, were excluded from the analysis.

Search terms for all fields were: drug-drug interaction, drug interaction, stem cell transplant, transplantation conditioning and conditioning regimen for PubMed; and drug interaction, stem cell transplantation and transplantation conditioning for Embase, combined with Boolean operators or and and. Additional publications cited in the selected articles were also included in the search given their significance in the authors’ opinion. Figure 1 shows the process followed to select the articles included in the search.

Figure 1.

Article selection process.

(0,14MB).

The most relevant data in each publication was extracted by one of the authors and independently checked by another author. Any discrepancies were resolved by a third investigator. The selected publications were then run through an initial filter where the title, abstract and main text of the articles were screened to ensure that they were in line with the purpose of the present study. When the same information was repeated, the most-up-to-date, clear and comprehensive study was selected.

The epidemiological features of all the studies were duly analyzed and recorded. Such features included: study type; database used; rate and prevalence of DDIs; overall number of DDIs; number of DDIs per patient; most usual DDIs; mechanism of action; severity: risk factors; and most commonly involved drugs; as well as the DDI effect observed in drug-to-drug studies.

The literature search resulted in 2,715 records (1,187 for PubMed and 1,528 for Embase). After removal of duplications, of articles whose title or abstract were unrelated to the purpose of this study, and of publications summarizing conference presentations, a total of 62 records were obtained that met the inclusion criteria. Ten of these were classified as epidemiological studies and 52 as studies analyzing the effects of DDIs. Tables 1 and 2 show the characteristics, objectives and main findings of the studies included.

Eight of the 10 epidemiological studies centered exclusively on patients undergoing HSCT, while the other two included hematological patients in general12,13. The study with the largest patient cohort had 384 subjects14. Six studies analyzed the DDIs resulting from all the drugs administered, while two studies were dedicated to DDIs associated with the use of antimicrobials15,16, one of them focusing on DDIs associated with anti-infectious, antineoplastic and immunomodulating drugs9, and the other on DDIs associated to mycophenolic acid17. Sánchez et al. conducted a separate study of DDIs with the potential to affect renal function, and associated them with a slower glomerular filtration rate18.

Most studies made specific mention of the database used, the prevalence of DDI, the drugs most commonly involved in DDIs, and the drug combinations most prone to DDIs, including those usually leading to the more severe forms of DDI. The most commonly used database was Micromedex®9,13,15,17. The study resulting in the highest number of overall DDIs was Gholaminezhad et al.14 with 13,600 DDIs in 384 patients, whereas the one detecting the lowest amount of DDIs was Guastaldi & Secoli16 with 13 DDIs in 35 patients.

The per-patient prevalence of DDIs ranged between the 60% reported by Guastaldi et al.19, who used the Drug Reax database® and the 100% found by Gholaminezhad et al.14, who used Lexi-Interact®. Using a different methodology, and in a hematologic cohort including more patients than just those undergoing HSCT, Fernández de Palencia et al.13 described a DDI prevalence per drug prescribed of 56.8% using the Drug Interaction Facts database® and of 74.1% using the Micromedex database®.

The mechanism of action of DDIs was mostly pharmacokinetic12,14,19. Interactions were mostly late-onset16,19 and the damage was mostly moderate14,16,19 or severe20. However, Sánchez et al. reported a lower incidence of pharmacokinetic than of pharmacodynamic DDIs18. Azole antifungals were among the drugs most commonly involved in DDIs in six of the 10 epidemiologic studies analyzed12–16,19.

According to Guastaldi & Secoli16, factors associated with the risk of developing DDIs included the number of drugs used12,14,16, the prescription of concomitant non-antineoplastic medication13,17, advanced age, and male sex.

Of the 52 studies analyzing the effect of DDIs, most of them pharmacokinetic, 18 focused on DDIs between azole antifungals and calcineurin inhibitors7,8,21–36; six on DDIs between azole antifungals and sirolimus37–42; five on DDIs between azole antifungals and other drugs (cyclophosphamide43, warfarin44, budesonide45, simvastatin6, and proton pump inhibitors46); 15 on DDIs between calcineurin inhibitors and other drugs (amphotericin B47, cyclophosphamide48, aprepitant49,50, micafungin51,52, caspofungin53, ritonavir54, calcium channel blockers55, imatinib56,57, fentanyl58, letermovir59,60, midostaurin61); five on DDIs between busulfan and other drugs (metronidazole62,63, N-acetylcysteine64, fludarabine65, blinatu-mumab66); one on DDIs between aprepitant and cyclophosphamide67; one on DDIs between rifampicin, sirolimus and voriconazole68; and one on DDIs between sirolimus and mirabegron69. The main effects of these DDIs are shown in table 2.

According to Yang et al.8, 10 patients experienced life-threatening complications potentially associated with DDIs between cyclosporine A and itraconazole or voriconazole on administration of supratherapeutic levels of cyclosporine. Six patients developed grade I to III graft-versus-host disease (GVHD) and eventually died from idiopathic pneumonia syndrome or alveolar hemorrhage. Another four patients died from neurological complications associated with cyclosporine A. Other authors have described serious adverse events such as rhabdomyolysis resulting from the interaction of azoles with statins6,7.

The present study was based on a compilation of studies reporting on the DDIs suffered by patients undergoing HSCT over the last 20 years. Two groups of studies can be distinguished: epidemiological studies on the one hand, and studies analyzing DDIs between two specific drugs or drug families on the other.

Epidemiological studies on the DDIs suffered by patients undergoing HSCT are scarce and tend to use dissimilar methodologies which lead to highly heterogeneous results. What can be concluded about them is that these studies show a high prevalence of DDIs, particularly potentially severe ones or between contraindicated medications. In addition, results tend to be rather heterogeneous on account, among other reasons, of the use of different databases.

Such heterogeneity can be easily seen in the literature70,71, where significant differences have been found in one same population when the data is analyzed using different databases. For example, Fernandez de Palencia et al.13 observed different prevalence rates in hematologic patients depending on whether they used the Lexi Interaction Facts® database (56.8%) or the Micromedex® database (74.1%). These differences were greater in the case of oncologic patients72 (80% using Micromedex® and 30% using Interaction Facts®) and smaller in pediatric hemato-oncologic patients73 (44.7% using Micromedex® and 51.3% using Drug Interaction Facts®). A statistical analysis of the concordances between the two databases, which compared the DDIs detected across 1,166 treatments, showed that concordance was weak in terms of the capacity to detect potential DDIs and nonexistent for the level of severity and scientific evidence attributed by each database to the same DDI71. This heterogeneity precludes the use of databases as decision-making tools in clinical practice.

Although epidemiological studies tend to focus on DDIs occurring between drugs that are commonly used in the context of HSCT and whose risk profile is well known, such as calcineurin inhibitors and azoles, other relevant yet somewhat less well-known DDIs such as those associated to CNS depressants (benzodiazepines, morphine derivatives, etc.), antiemetics, corticosteroids, proton pump inhibitors, antidepressants or antibiotics are also reported, albeit in other scenarios74–76.

Other studies, dealing mostly with DDIs between two agents, usually focus on pharmacokinetic mechanisms of action, with none of them to the best of our knowledge analyzing pharmacodynamic DDIs. All pharmacokinetic studies exhibit a similar data collection pattern.

Most published articles describe DDIs between azole antifungals and calcineurin inhibitors, which result in increased serum concentrations of the calcineurin inhibitors caused by an inhibition of the CYP3A4 metabolism. However, the intensity of DDIs tends to depend on the specific antifungal or calcineurin inhibitor used. For example, the increase in calcineurin inhibitor serum concentrations is apparently greater with voriconazole than with fluconazole22,35, whereas the intensity of DDIs involving itraconazole is significantly higher with tacrolimus than with cyclosporine28. Several studies have shown the level of azole concentration not to be associated with an increase in calcineurin inhibitor concentrations26,27. Masoumi et al., however, did find a correlation in that respect31. The route of administration also seems to play an important role. Indeed, fluconazole tends to result in more significant DDIs with intravenously administered calcineurin inhibitors when it is administered orally rather than intravenously23. In the same vein, orally administered tacrolimus is more affected by concomitant use of voriconazole36. It must be pointed out that one of the studies analyzed claims that a DDI between cyclosporine and voriconazole or itraconazole was potentially responsible for the death of 10 patients who experienced subtherapeutic levels of cyclosporine. Six of them died as a result of an idiopathic pneumonia syndrome or an alveolar hemorrhage following the occurrence of GVHD and four as a result of neurologic complications associated to cyclosporine A8.

Azole antifungals are usually implicated in a different class of DDI when used together with sirolimus37–42. This combination leads to increased serum concentrations of sirolimus, requiring an empiric dose reduction before it can be combined with azoles. Antifungals may also interact with agents like cyclophosphamide, with fluconazole being safer than itraconazole as the former inhibits CYP2C9, which leads to lower levels of 4-hydroxycyclo-phosphamide, a toxic cyclophosphamide metabolite43. Azoles can also increase serum concentrations of warfarin44, budesonide45, simvastatin6, and lansoprazole46.

Calcineurin inhibitors, for their part, are involved in a significant number of DDIs with agents other than azoles. Tacrolimus and cyclosporine concentrations do not seem to be influenced by the presence of micafungin51,52, and although caspofungin does not seem to have an influence on tacrolimus concentration levels, but it does seem to increase serum concentrations of cyclosporine53. When combined with amphotericin B, cyclosporine leads to a statistically significant worsening of renal function, which may be clinically controlled if the drug is infused over a 24-hour period and if strict salt replenishment is observed47. Cyclosporine has also been found to be involved in DDIs with cyclophosphamide, leading to a decrease in the latter's serum concentrations48.

There is generalized concurrence in the literature regarding the effects of DDIs occurring between calcineurin inhibitors and other drugs such as imatinib, which leads to increased cyclosporine concentrations56,57. Nonetheless, in a study of 85 patients, Shayani et al.50 concluded that aprepitant increases sirolimus but not tacrolimus serum concentrations. Conversely, a study of 26 patients by Ibrahim et al.49 found that aprepitant did increase tacrolimus serum concentrations. Another example is provided by a study of 46 patients by Maples et al.60, who claimed that it is not necessary to adjust the dose of calcineurin inhibitors prior to administration of letermovir. However, a 3-case series published by Guo et al.59 showed that letermovir inhibits CYP3A4 and increases tacrolimus concentrations, which according to these authors warrants a dose reduction.

Some studies have shown that agents belonging to the same therapeutic group can have different DDI profiles. In this respect, Bernard et al.55 found that nicardipine and amlodipine increase serum concentrations of cyclosporine, while lacidipine had no effect of such concentrations. Finally, two studies showed that concomitant administration of metronidazole62 and fludarabine65 with busulfan increases the latter's serum concentrations.

In 2008, Vives et al. published a series of case reports that brought to light a series of severe adverse events associated with some DDIs. The authors reported that concomitant use of simvastatin, cyclosporine A and risperidone can result in rhabdomyolysis and renal failure7. Moreover, in a study describing significant DDIs resulting from recently introduced agents, Mancini et al. reported a 70% increase in the serum levels of cyclosporine A when used together with midostaurin61.

The main limitation of the present study is associated with the heterogeneity of the studies analyzed and the absence of data to quantify the impact of DDIs on patient outcomes.

In summary, all the studies analyzed describe a high prevalence of DDIs in patients undergoing HSCT, with significant disparities across the different authors in terms of the prevalence and characteristics of the DDIs identified. Such disparities are attributable to the way the studies were designed and the databases used. Factors related to the risk of DDIs include the number of drugs concomitantly administered. All studies on DDIs are of a pharmacokinetic nature and focus mainly on DDIs between azole antifungals and calcineurin inhibitors, or between these two drug families and other agents. It would be important to unify the criteria followed by epidemiological studies to produce more consistent outcomes conducive to the implementation of effective risk reduction strategies. It would also be essential to investigate pharmacodynamic DDIs and to carry out a more in-depth analysis of DDIs between other commonly used drugs in the HSCT context and between drugs that have been recently introduced in our therapeutic arsenal.

No funding.

No conflict of interests.