Abstract
Dasatinib absorption is pH-dependent, and concomitant use of Proton Pump Inhibitors (PPIs) can substantially reduce dasatinib exposure. Whether this interaction has clinical relevance and affects treatment outcomes in patients with chronic myeloid leukemia (CML) remains unclear. This retrospective cohort study used German statutory health insurance claims data (2010 − 2024). Adults with CML who initiated dasatinib treatment were selected. Patients were classified according to concomitant PPI use around dasatinib initiation. The primary endpoint was time to switch to another systemic CML therapy. Two analytical cohorts were evaluated (incident and prevalent cohorts). Propensity score matching and Cox proportional hazards models were used to compare patients with and without PPI co-medication. Sensitivity analyses censored follow-up at severe adverse events and restricted the population to users of the originator dasatinib formulation. Among 297 dasatinib patients in the incident and 372 in the prevalent cohort, PPI co-medication was observed in approximately 19–21% of patients. After matching, concomitant PPI use was consistently associated with earlier treatment switching. This association reached statistical significance in the prevalent cohort (hazard ratio: 1.87, 95% CI 1.18–2.95; p = .008) and was directionally consistent in the incident cohort. Sensitivity analyses yielded directionally consistent findings. In this large real-world analysis, concomitant PPI use was associated with earlier treatment switching in dasatinib-treated patients with CML. While causality cannot be established, these findings are consistent with the known pH-dependent interaction and reinforce existing recommendations to avoid PPI co-medication during dasatinib therapy whenever feasible.
Supplementary Information
The online version contains supplementary material available at 10.1007/s00277-026-07211-w.
Keywords: Chronic myeloid leukemia, Tyrosine kinase inhibitors, Proton pump inhibitors, Drug interactions, Pharmacoepidemiology
Introduction
Chronic Myeloid Leukemia (CML) is a rare hematologic malignancy characterized by the presence of the BCR::ABL1 fusion gene, resulting in continuous tyrosine kinase activity and uncontrolled myeloid proliferation [1]. The introduction of BCR::ABL1-targeted tyrosine kinase inhibitors (TKIs) has transformed CML from a fatal disease into a manageable chronic condition with near-normal life expectancy for many patients [2, 3]. As survival improves, patients increasingly require long-term TKI therapy alongside treatment for comorbidities, making clinically relevant drug–drug interactions an important consideration in routine care.
Dasatinib is a second-generation TKI approved and widely used across different lines of therapy in CML [4]. As dasatinib absorption is pH-dependent, concomitant use of Proton Pump Inhibitors (PPIs) may substantially reduce drug exposure [5, 6]. Pharmacokinetic studies have reported reductions in dasatinib plasma exposure exceeding 40% – 60% during acid-suppressive therapy, particularly for crystalline formulations [5, 6]. Despite recommendations to avoid concomitant PPI use during dasatinib treatment, PPIs remain commonly prescribed among patients with cancer due to gastrointestinal symptoms, polypharmacy, and age-related comorbidities [7–10].
Although the pharmacokinetic interaction between dasatinib and PPIs is well established, evidence regarding the clinical consequences remains limited [7]. Previous real-world studies have reported poorer survival outcomes and increased treatment discontinuation in oncologic patients receiving TKIs with concomitant PPIs [8, 9, 11]. However, evidence regarding treatment switching among dasatinib-treated patients with CML remains limited, particularly in Germany. As direct measures of treatment response are generally unavailable in administrative claims data [12], treatment switching may serve as a clinically meaningful real-world indicator of treatment outcomes [13].
Therefore, this study aimed to evaluate the association between PPI co-medication and treatment switching among dasatinib-treated patients with CML using German statutory health insurance (SHI) claims data.
Methods
Data source
This non-interventional retrospective cohort study used anonymized German SHI claims data provided by the GWQ ServicePlus AG, covering more than 7 million insured individuals (approximately 10% of the German SHI population). Available data include inpatient and outpatient diagnoses, procedures, prescription dispensations, healthcare utilization, demographic information, and mortality. Diagnoses, procedures, and medications are coded using the German Modification of the International Statistical Classification of Diseases and Related Health Problems, 10th Revision (ICD-10-GM), the German Procedure Classification (Operationen- und Prozedurenschlüssel; OPS), and the Anatomical Therapeutic Chemical (ATC) coding system, respectively. Data from 2010 to 2024 were available for cohort identification and follow-up.
Study population
Individuals were selected from anonymized German SHI claims data between January 1, 2011, and September 30, 2024, allowing a 12-month baseline period before dasatinib initiation and at least 90 days of potential follow-up thereafter (Fig. 1). Eligible individuals had at least one confirmed outpatient or one inpatient diagnosis consistent with CML (ICD-10-GM: C92*) and a newly observed dasatinib treatment episode identified through inpatient procedure codes (OPS 6-004.3*) or outpatient prescription claims (ATC L01XE06 or L01EA02). The broader C92* category was used to account for variation in coding specificity within routine claims data. The first observed dasatinib treatment episode was defined as the index date. Individuals were required to be aged ≥ 18 years, continuously insured for at least 12 months before the index date, and have no evidence of dasatinib treatment during the baseline period. Individuals with other primary malignancies (ICD-10-GM: C00–C97, excluding leukemia codes C91–C95) during baseline or with less than 90 days of available follow-up after the index date were excluded.
Fig. 1.
Study design and assessment windows for the main and expanded analysis samples. Abbreviations: CML, chronic myeloid leukemia; C92*, ICD-10-GM diagnosis category for CML or related myeloid leukemia; Dasa + PPI, dasatinib with proton pump inhibitor co-medication; Dasa-only, dasatinib without proton pump inhibitor co-medication; LOT, line of therapy; PPI, proton pump inhibitor
Two analysis samples were constructed. The main analysis sample (incident cohort) included individuals without a recorded C92* diagnosis during the 12 months preceding the first qualifying diagnosis, allowing assessment of treatment sequencing and line of therapy at dasatinib initiation (Fig. 1 – Panel A). The expanded analysis sample (prevalent cohort) included all eligible individuals irrespective of prior C92* diagnoses, thereby increasing sample size but precluding reliable reconstruction of prior treatment history and line of therapy (Fig. 1 – Panel B).
Within each analysis sample, individuals were classified according to PPI use identified from outpatient prescription claims (ATC A02BC*) around dasatinib initiation. In the incident cohort, a narrower PPI exposure definition was applied using an exposure assessment window from 60 days before to 30 days after the index date [-60,+30]. Individuals were classified as Dasa + PPI if PPI use was observed within 30 days after dasatinib initiation, or if a pre-index PPI prescription was followed by a continued PPI treatment during the first 6 months after initiation. All remaining individuals were classified as Dasa-only. In the prevalent cohort, a broader PPI exposure definition was used by extending the post-index exposure assessment window to 60 days after dasatinib initiation [-60,+60]. This approach was intended to increase sensitivity for identifying PPI co-medication in routine clinical practice and reduce potential exposure misclassification.
Population adjustment
Propensity score matching (PSM) was applied separately within each cohort using logistic regression models including age, sex, index year, and comorbidity burden. In the incident cohort, the treatment line at dasatinib initiation was additionally included. Nearest-neighbor matching without replacement was performed using caliper width of 0.2 times the standard deviation (SD) of the logit of the propensity score. Individuals with PPI co-medication were matched 1:1 in the incident cohort and 1:2 in the prevalent cohort. Covariate balance was assessed using standardized differences, mean bias and Rubin’s B statistic.
Outcome assessment
Patient characteristics were assessed during the 12-month pre-index period or at the index date, as applicable. These included age, sex, qualifying diagnosis subcode, prior TKI use, prior chemotherapy, and comorbidity burden assessed using the Charlson Comorbidity Index (CCI) and Elixhauser Comorbidity Index [14, 15]. In the incident cohort, CML disease duration was estimated as the time between the first observed qualifying diagnosis and dasatinib initiation. The line of therapy at dasatinib initiation was additionally assessed during this period.
Follow-up began on the day after dasatinib initiation and continued until the earliest occurrence of death, switch to another systemic CML therapy, discontinuation of dasatinib without treatment switch, change in PPI exposure status, or censoring due to the end of data availability (Fig. 1). For individuals in the Dasa + PPI group, follow-up was censored at discontinuation of PPI treatment. For individuals in the Dasa-only group, follow-up was censored at initiation of PPI treatment after cohort entry. Discontinuation of dasatinib or PPI treatment was defined as a gap exceeding the estimated days of supply derived from defined daily doses after the last prescription or administration. Exposure-based censoring was applied to minimize exposure misclassification.
The primary endpoint was time to treatment switch, defined as the number of days from the index date to initiation of another systemic therapy for CML. Treatment switch therapies included alternative TKIs and chemotherapy identified from inpatient procedures and outpatient prescription claims. Evaluated TKIs included imatinib, nilotinib, bosutinib, ponatinib, and asciminib. The ATC and OPS codes used to define treatment switch therapies are provided in Supplementary Table S1.
Two sensitivity analyses were conducted. First, follow-up was censored at the occurrence of a severe adverse event (SAE), defined as a hospitalization with a relevant main diagnosis. Evaluated SAE categories included respiratory complications, myelosuppression-related events, and gastrointestinal disorders associated with dasatinib treatment. ICD-10-GM codes are provided in Supplementary Table S2.
Second, analyses were restricted to patients receiving the originator dasatinib product Sprycel®, identified using Pharmaceutical Central Numbers (Pharmazentralnummern; PZN) codes (Supplementary Table S3). Both sensitivity analyses used the same exposure definitions, matching procedures, and outcome assessment as the primary analysis.
Statistical analysis
All statistical analyses were conducted using Stata version 15 (StataCorp, College Station, TX, USA). Baseline characteristics were summarized descriptively. Time-to-event analyses were conducted using Kaplan–Meier (KM) methodology and Cox proportional hazards models. Hazard ratios (HRs) and 95% CIs comparing exposure groups were estimated in the matched cohorts.
Results
Study cohort selection
Between January 1, 2011, and September 30, 2024, a total of 526 patients with at least one qualifying diagnosis consistent with CML (ICD-10-GM: C92*) and evidence of dasatinib treatment were identified in the GWQ ServicePlus database. After applying inclusion and exclusion criteria, 372 patients were included in the prevalent cohort, and 297 patients met the additional incident criterion (Fig. 2).
Fig. 2.
Cohort selection flowcharts of the main and expanded analysis samples. Abbreviation: ICD-10-GM = International Classification of Diseases, 10th Revision, German Modification; Dasa + PPI = dasatinib with PPI comedication; Dasa-only = dasatinib without PPI comedication; PPI = proton pump inhibitor
In the incident cohort (main analysis sample), 56 patients (18.9%; 56/297) were classified into the Dasa + PPI group and 241 (81.1%; 241/297) into the Dasa-only group before matching. Among patients in the Dasa + PPI group, 32 (57.1%) had evidence of PPI treatment before dasatinib initiation, whereas the remaining patients initiated PPI therapy after starting dasatinib. Among patients initially classified into the Dasa-only group, subsequent initiation of PPI therapy later during follow-up was common. Overall, only 35.0% of patients in the incident cohort had no observed PPI use throughout follow-up. In the sensitivity analysis, restricted to patients receiving the originator dasatinib product Sprycel®, 176 patients (27 with PPI comedication and 149 without comedication) were identified.
In the prevalent cohort (expanded analysis sample), 77 patients (20.7%; 77/372) were classified into the Dasa + PPI group and 295 (79.3%; 295/372) into the Dasa-only group before PSM according to the cohort-specific PPI exposure definitions (Sprycel® subgroup: 46 with PPI comedication and 191 without PPI comedication).
After PSM, the incident cohort included 56 matched patients per exposure group, whereas the prevalent cohort included 65 patients with PPI co-medication matched to 130 patients without PPI co-medication.
Patient characteristic
Patient characteristics before and after PSM are presented in Table 1 for the incident cohort and Table 2 for the prevalent cohort. Before matching, patients with PPI comedication were generally older and had a higher comorbidity burden than patients without PPI comedication in both cohorts. Most patients had a qualifying diagnosis of C92.1 (BCR::ABL1-positive CML) at cohort entry, whereas the remaining patients were identified based on other qualifying C92 diagnosis codes together with the evidence of dasatinib in the study definition. In the incident cohort, approximately one-third of patients in each exposure group received dasatinib as first-line therapy, one-third as second-line therapy, and one-third as third- or later-line therapy (Table 1). The estimated initial prescribed outpatient dasatinib dose during the first 6 months after treatment initiation was similar between groups (mean dose: 99.7 mg/day in the Dasa + PPI group and 99.8 mg/day in the Dasa-only group). In the prevalent cohort, prior TKI exposure before dasatinib treatment was common (46.8% in the Dasa + PPI group and 54.2% in the Dasa-only group before matching; Table 2). After PSM, baseline characteristics were well balanced between exposure groups in both cohorts, with mean bias below 5% and Rubin’s B values below 25. Patient characteristics among Sprycel® users before and after matching in the sensitivity analyses are provided in the Supplementary Table S4 and Table S5.
Table 1.
Patient characteristics of the incident (main analysis) sample before and after propensity score matching
| Before matching | After matching | ||||
|---|---|---|---|---|---|
| Dasa + PPI (n = 56) |
Dasa-only (n = 241) |
Dasa + PPI (n = 56) |
Dasa-only (n = 56) |
||
| Demographics | |||||
| Median age (mean ± SD) | 59 (58.5 ± 13.5) | 52 (52.0 ± 15.2) | 59 (58.5 ± 13.5) | 58.5 (58.0 ± 15.9) | |
| Female, n (%) | 23 (41.1) | 92 (38.2) | 23 (41.1) | 24 (42.9) | |
| Clinical characteristics | |||||
| C92.1 CML, BCR::ABL1-positive, n (%) | 54 (96.4) | 221 (91.7) | 54 (96.4) | 53 (94.6) | |
|
Median CML duration in days (mean ± SD) |
34 (316 ± 711) | 89 (463 ± 745) | 34 (316 ± 711) | 105 (542 ± 920) | |
| LOT of dasatinib, n (%) | |||||
| 1st LOT | 20 (35.7) | 80 (33.2) | 20 (35.7) | 19 (33.9) | |
| 2nd LOT | 19 (33.9) | 76 (31.5) | 19 (33.9) | 18 (32.1) | |
| 3rd LOT+ | 17 (30.4) | 85 (35.3) | 17 (30.4) | 19 (33.9) | |
| Prior TKI use, n (%) | 22 (39.3) | 117 (48.5) | 22 (39.3) | 30 (53.6) | |
| Prior chemotherapy, n (%) | 25 (44.6) | 96 (39.8) | 25 (44.6) | 20 (35.7) | |
| Comorbidity profile | |||||
| Median CCI (mean ± SD) | 3 (3.8 ± 2.2) | 2 (2.8 ± 1.3) | 3 (3.8 ± 2.2) | 3 (3.3 ± 1.7) | |
|
Median Elixhauser index (mean ± SD) |
2 (4.9 ± 8.9) | 2 (3.9 ± 6.5) | 2 (4.9 ± 8.9) | 3 (5.5 ± 8.3) | |
Note: Index date = the date of the first observed dasatinib treatment; prior chemotherapy was defined as receipt of chemotherapy between the initial CML diagnosis and dasatinib initiation, identified from inpatient procedure codes or outpatient prescriptions. This included hydroxycarbamide (hydroxyurea), cytarabine, daunorubicin, cytarabine/daunorubicin combinations, purine analogues, vincristine, and unspecified inpatient chemotherapy procedures
Abbreviation: CCI = Charlson comorbidity index; CML = chronic myeloid leukemia; Dasa + PPI = dasatinib with PPI comedication; Dasa-only = dasatinib without PPI comedication; LOT = line of therapy; PPI = proton pump inhibitor; SD = standard deviation; TKI = tyrosine kinase inhibitor
Table 2.
Patient characteristics of the prevalent (expanded analysis) sample before and after propensity score matching
| Before matching | After matching | ||||
|---|---|---|---|---|---|
| Dasa + PPI (n = 77) |
Dasa-only (n = 295) |
Dasa + PPI (n = 65) |
Dasa-only (n = 130) |
||
| Demographics at index | |||||
| Median age (mean ± SD) | 58 (57.6 ± 15.0) | 53 (53.0 ± 14.5) | 56 (55.3 ± 14.8) | 56.5 (55.5 ± 13.7) | |
| Female, n (%) | 34 (44.2) | 110 (37.3) | 29 (44.6) | 57 (43.8) | |
| Clinical characteristics | |||||
| C92.1 CML, BCR::ABL1-positive, n (%) | 71 (92.2) | 275 (93.2) | 59 (90.8) | 119 (91.5) | |
| Prior TKI use, n (%) | 36 (46.8) | 160 (54.2) | 31 (47.7) | 71 (54.6) | |
| Prior chemotherapy, n (%) | 30 (39.0) | 98 (33.2) | 26 (40.0) | 40 (30.8) | |
| Comorbidity profile | |||||
| Median CCI (mean ± SD) | 3 (3.7 ± 2.1) | 2 (2.9 ± 1.4) | 3 (3.0 ± 1.4) | 2 (3.0 ± 1.4) | |
|
Median Elixhauser index (mean ± SD) |
2 (4.0 ± 8.1) | 0 (3.8 ± 6.4) | 0 (3.2 ± 7.8) | 0 (3.2 ± 6.0) | |
Note: Index date = the date of the first observed dasatinib treatment; CML duration until dasatinib initiation was not assessed in the expanded analysis sample because no assumption of incident CML diagnosis was applied in this sample; prior chemotherapy was defined as receipt of chemotherapy prior to the dasatinib initiation, identified from inpatient procedure codes or outpatient prescriptions. This included hydroxycarbamide (hydroxyurea), cytarabine, daunorubicin, cytarabine/daunorubicin combinations, purine analogues, vincristine, and unspecified inpatient chemotherapy procedures
Abbreviation: CCI = Charlson comorbidity index; CML = chronic myeloid leukemia; Dasa + PPI = dasatinib with PPI comedication; Dasa-only = dasatinib without PPI comedication; PPI = proton pump inhibitor; SD = standard deviation; TKI = tyrosine kinase inhibitor
Time to treatment switch
KM estimates for the time to treatment switch are presented in Fig. 3.
Fig. 3.
Kaplan–Meier curves for time to treatment switch among dasatinib-treated patients with chronic myeloid leukemia (CML) according to proton pump inhibitor (PPI) co-medication status. Panel A shows the primary analysis in the incident cohort. Panel B shows the primary analysis in the prevalent cohort. Panel C shows the sensitivity analysis in the incident cohort censoring follow-up at the first severe adverse event (SAE). Panel D shows the corresponding sensitivity analysis in the prevalent cohort. Note: Treatment switch was defined as initiation of another systemic CML therapy. Cell counts < 5 in the number-at-risk table and failure counts were censored in accordance with data protection requirements. Abbreviations: CI = confidence interval; CML = chronic myeloid leukemia; Dasa + PPI = dasatinib with PPI comedication; Dasa-only = dasatinib without PPI comedication; HR = hazard ratio; PPI = proton pump inhibitor; SAE = severe adverse event
In the matched incident cohort, treatment switch occurred in 28 of 56 patients (50.0%) in the Dasa + PPI group and in 23 of 56 patients (41.1%) in the Dasa-only group. Median time to treatment switch was 16.1 months (95% CI 9.5–40.7) in the Dasa + PPI group and 18.6 months (95% CI 13.2–83.3) in the Dasa-only group. Although not statistically significant, PPI co-medication was associated with a trend toward earlier treatment switching (HR 1.58, 95% CI 0.89–2.78; p=.117).
In the matched prevalent cohort, treatment switch to another systemic CML therapy occurred in 31 of 65 patients (47.7%) in the Dasa + PPI group and in 47 of 130 patients (36.2%) in the Dasa-only group. Median time to treatment switch was shorter among patients with PPI co-medication than among patients without PPI co-medication (16.1 months [95% CI 7.2–40.7] vs. 38.1 months [95% CI 30.7–68.6]). PPI co-medication was associated with a significantly higher hazard of early treatment switch (HR 1.87, 95% CI 1.18–2.95; p=.008).
In the sensitivity analyses censoring follow-up at the first SAE, treatment switch occurred in 28 out of 65 patients (43.1%) in the Dasa + PPI group versus 42 out of 130 (32.3%) patients in the Dasa-only group in the matched prevalent cohort, and 25 out of 56 patients (44.6%) versus 19 out of 56 patients (33.9%) in the matched incident cohort. Median time to treatment switch censoring follow-up at the first SAE was shorter among patients with PPI co-medication than among patients without PPI co-medication in both cohorts (17.0 months [95% CI 7.2–not reached] vs. 39.4 months [95% CI 37.1–not reached] in the matched prevalent cohort; 17.0 months [95% CI 9.5–not reached] vs. 66.9 months [95% CI 15.9–not reached] in the matched incident cohort). Findings remained directionally consistent with the primary analysis in both cohorts. In the prevalent cohort, PPI co-medication remained associated with a significantly higher hazard of earlier treatment switch (HR 1.88, 95% CI 1.16–3.05; p=.011), whereas the incident cohort showed a numerically increased but non-significant hazard estimate (HR 1.70, 95% CI 0.91–3.15; p=.095) (Fig. 3).
Sensitivity analyses restricted to the originator dasatinib formulation (Sprycel®) showed directionally consistent findings in the prevalent cohort. Results in the incident cohort were inconclusive due to the substantially reduced sample size (27 patients per group) and overlapping KM curves (Supplementary Figure S1).
Discussion
Principle findings
In this retrospective real-world evidence study using German SHI claims data, concomitant PPI use among dasatinib-treated patients with CML was associated with earlier treatment switching compared with dasatinib treatment without PPI co-medication. The association was strongest in the prevalent cohort, which included a larger population of dasatinib-treated patients and applied a broader PPI exposure definition. In this sample, PPI co-medication was associated with a shorter median time to treatment switch and a significantly higher hazard of switching to another systemic CML therapy. In the incident cohort, effect estimates were directionally consistent but did not reach statistical significance, possibly reflecting the smaller sample size and reduced statistical precision. Findings remained directionally consistent in sensitivity analyses designed to reduce the influence of severe toxicity-related treatment changes and to evaluate patients receiving the originator crystalline dasatinib formulation Sprycel®.
As direct measures of treatment response and physician-reported reasons for treatment modification are not available in administrative claims data, treatment switching was used as a clinically relevant real-world endpoint. Although treatment switching may reflect inadequate response, loss of response, intolerance, toxicity, disease progression, or other clinical considerations [16], it may also reflect physician recognition of the clinically relevant interaction between dasatinib and PPIs and a proactive decision to switch therapy when acid-suppressive treatment could not be discontinued. Although sensitivity analyses censoring follow-up at SAEs suggests that the observed association was unlikely to be explained solely by severe toxicity- or intolerance-related treatment changes, the underlying clinical reason for treatment switching could not be determined in the available data.
Comparison with previous literature
The findings of this study are consistent with pharmacokinetic evidence demonstrating that gastric acid suppression reduces dasatinib exposure due to its pH-dependent absorption characteristics [5–7]. Previous studies reported reductions in dasatinib plasma exposure exceeding 40%–60% during concomitant acid-suppressive therapy, particularly for crystalline formulations of dasatinib [5, 6]. Although the pharmacokinetic interaction is well established, evidence regarding its clinical consequences has remained limited [7]. The directionally consistent findings observed in analyses restricted to the originator formulation Sprycel® further support the biological plausibility of the observed association, as this formulation has been shown to be particularly susceptible to acid-dependent reductions in absorption [5, 11, 17].
However, not all clinical studies have reported detrimental effects of concomitant acid-suppressive therapy. Haddad et al. observed no significant differences in major molecular responses between patients receiving concomitant PPIs and dasatinib and those receiving dasatinib alone. Notably, patients in that study received a reduced starting dose of dasatinib (50 mg/day), and the primary outcomes focused on molecular response rather than treatment switching [18]. Similarly, Koutake et al. reported lower rates of dasatinib discontinuation among patients receiving concomitant PPIs or H2-receptor antagonists. However, only five patients in that study received concomitant PPI and dasatinib, substantially limiting the precision and interpretability of the findings for PPI users specifically [19]. These differences in study design, including sample sizes, dasatinib dosing, exposure definitions and clinical endpoints, may contribute to the differing findings across studies. Whereas these two studies primarily evaluated molecular response or treatment discontinuation, the present study examined treatment switching as a clinically relevant real-world endpoint using nationwide administrative claims data.
Several previous observational studies reported poorer survival outcomes and increased treatment discontinuation among patients receiving concomitant TKIs and PPIs across oncology populations [8, 9, 11]. Sharma et al. observed that concomitant use of TKIs and PPIs was associated with worse survival outcomes and earlier discontinuation of therapy among older adults with cancer, highlighting the potential clinical consequences and healthcare burden associated with reduced TKI effectiveness [8]. Our findings extend this evidence by specifically evaluating treatment-switching patterns among dasatinib-treated patients with CML in German routine care.
Unlike previous German claims-based CML studies that restricted cohort identification to selected specific ICD-10-GM subcodes, such as C92.1, C92.2, C92.7, and C92.9 [20], this study used the broader C92* category in combination with dasatinib treatment. Accordingly, the qualifying diagnosis at cohort entry was not always C92.1 (BCR::ABL1-positive CML), as the fourth-digit ICD-10-GM subcode may not be recorded consistently in routine clinical practice. This approach was intended to account for variability in coding practices within routine claims data, including instances in which providers may not consistently document the fourth-digit subcode. As dasatinib is predominantly used for BCR::ABL1-positive CML, the combined use of C92* diagnoses and dasatinib treatment was considered a pragmatic approach to maximize cohort completeness while maintaining clinical relevance.
In the incident cohort, more than half of patients with PPI co-medication (57.1%) had evidence of PPI treatment before initiating dasatinib, whereas the remaining patients initiated PPI therapy after dasatinib treatment had started. This suggests that concomitant PPI use frequently reflected pre-existing acid-suppressive therapy rather than therapy initiated after dasatinib initiation. Consequently, opportunities to avoid this clinically relevant drug–drug interaction may exist both at the time of dasatinib initiation and during subsequent treatment.
Our findings should also be considered within the context of real-world prescribing practices. Recent studies have reported that TKIs and PPIs were frequently prescribed by different physicians, increasing the risk that clinically relevant drug–drug interactions remain unrecognized [10, 11]. Consistent with this observation, a substantial proportion of patients initially classified as Dasa-only subsequently initiated PPI therapy during follow-up. This highlights the dynamic nature of acid-suppressive therapy use in routine care and suggests that a single baseline assessment of PPI exposure may not adequately capture ongoing exposure patterns among dasatinib-treated patients.
Patients receiving concomitant PPI therapy also tended to have a higher comorbidity burden before matching, which may reflect greater overall healthcare utilization and a higher likelihood of concomitant medication use. Because polypharmacy is common among patients with multiple comorbidities, it may contribute to treatment complexity and increase the potential for clinically relevant drug–drug interactions. However, age, comorbidity burden, and several other baseline characteristics were balanced after propensity score matching, suggesting that these measured differences are unlikely to fully explain the observed association. Nevertheless, residual confounding from unmeasured clinical characteristics and concomitant medications cannot be completely excluded.
Beyond prescribing complexity, the findings should also be interpreted within the broader context of CML treatment sequencing. In the incident cohort, dasatinib was used across multiple lines of therapy rather than predominantly in heavily pretreated patients, suggesting that the observed association was not limited to advanced or highly refractory disease settings. This observation is consistent with previous real-world studies demonstrating substantial early treatment modifications and interruptions among throughout the CML treatment pathway [9].
Clinical and policy implications
Given the high cost of TKIs and the long-term treatment duration commonly required in CML management, reduced effectiveness associated with avoidable drug–drug interactions could contribute to unnecessary healthcare expenditures and increased treatment burden [8]. Earlier treatment switching may also result in additional monitoring, treatment escalation, and downstream healthcare utilization.
The findings reinforce existing recommendations to avoid concomitant PPI use during dasatinib treatment whenever clinically feasible [8]. Given the frequent involvement of different prescribers in TKI and PPI prescribing pathways [10, 11], medication reconciliation, interdisciplinary communication, and patient counseling to increase the awareness of medication safety regarding prescription and over-the-counter acid-suppressive therapy remain important [11].
Strengths and limitations
This study has several strengths. First, the use of a large longitudinal German claims database enabled assessment of real-world treatment patterns across inpatient and outpatient care settings over an extended observation period. Although the database does not represent a random sample of the German SHI population, its demographic composition is broadly comparable to that of the German SHI population overall, supporting the generalizability of descriptive findings. Second, the use of standardized coding systems and pre-specified exposure definitions, endpoints, censoring strategies, and analysis procedures supported methodological transparency and reproducibility. Third, the inclusion of both incident and prevalent analytical cohorts, together with PSM and multiple sensitivity analyses, allowed evaluation of the robustness of findings across different patient populations, exposure definitions, and analytical assumptions.
Several limitations should also be considered. Administrative claims data do not contain direct clinical measures of treatment response, such as cytogenetic and molecular responses, nor physician-reported reasons for treatment initiation or modification. Consequently, it was not possible to determine whether dasatinib was initiated because of intolerance, treatment failure, disease progression, or other clinical considerations following prior TKI therapy. Likewise, treatment switching could not be attributed to a specific underlying cause and may reflect a combination of inadequate response, disease progression, toxicity, intolerance, physician preference, or other clinical considerations [16]. Although sensitivity analyses censoring follow-up at hospitalized adverse events were conducted, non-hospitalized adverse events and lower-grade toxicities that may also contribute to treatment switching could not be captured. Exposure misclassification is also possible because outpatient medication use was inferred from dispensing claims rather than actual medication intake, and over-the-counter PPI use is not captured in German claims data. Furthermore, despite PSM, residual confounding cannot be excluded. Although matching improved the balance of most observed baseline characteristics, differences in claims-based CML disease duration remained between exposure groups, with a shorter median time from first observed CML diagnosis to dasatinib initiation among patients receiving PPI co-medication. This may reflect differences in treatment pathways or underlying clinical characteristics that were not fully captured in the available data. Moreover, important clinical indicators of disease severity and prognosis, including ELTS and Sokal risk scores, were not available in the claims data and therefore could not be considered in the matching or adjusted analyses [21]. A direct measure of polypharmacy was also not included, and residual confounding from differences in concomitant medication burden cannot be excluded. In addition, claims-based disease duration reflects the time since the first observed diagnosis within the available data period and may not represent the true duration of disease because of left-censoring. Finally, the relatively small sample sizes in some subgroups and sensitivity analyses limited statistical precision and precluded robust stratified analyses for selected clinically relevant subgroups, such as treatment line.
Conclusion
In this German real-world evidence study based on SHI claims data, concomitant PPI use among dasatinib-treated patients with CML was associated with earlier switching to another systemic CML therapy. Although claims data do not permit direct assessment of treatment response or the reasons underlying treatment modification, the observed association was consistent across multiple analytical approaches. These findings suggest that concomitant PPI use may contribute to earlier treatment modification in routine CML care and support existing recommendations to carefully review the need for acid-suppressive co-medication during dasatinib treatment.
Supplementary Information
Below is the link to the electronic supplementary material.
Acknowledgements
The authors would like to thank Antje Mevius, Research Principal at Institut für Pharmakoökonomie und Arzneimittellogistik e.V. (IPAM), for her expert guidance regarding the use and interpretation of German statutory health insurance claims data throughout the conduct of this study.
Author contributions
All authors contributed to the study’s conception and design. SM and CP performed data analysis and wrote the first draft of the manuscript, and all authors commented on previous versions. All authors read and approved the final manuscript and agree to be accountable for all aspects of the work.
Funding
This study was sponsored by Zentiva, a.s., Prague, Czech Republic. GIPAM GmbH was contracted by Zentiva for the conduct of this study.
Data availability
The datasets generated and/or analyzed during the current study are not publicly available since the findings of this study are extracted from health insurance claims. Data were available for research purposes from the sickness fund upon request, in an anonymized form. Due to restrictions around revealing patients’ confidential information, data were used under license for the current study, and so are neither publicly available nor can be shared further.
Declarations
Ethics approval
This study was a non-interventional, retrospective analysis of anonymized secondary data derived from German statutory health insurance claims. In accordance with applicable German data protection regulations, only anonymized data were available to the investigators, and no direct patient contact occurred. The study was conducted in accordance with the ethical principles of the Declaration of Helsinki and applicable regulations governing the use of anonymized secondary healthcare data.
Consent to participate
Informed consent was not required for this study because it was based exclusively on anonymized German statutory health insurance claims data. No direct contact with patients occurred, and no identifiable personal data were available to the investigators.
Competing interest
This study was sponsored by Zentiva, a.s., Prague, Czech Republic. Authors JB, JH, and FL are employees of Zentiva. CP is an employee of GIPAM GmbH. SM and TW are shareholders of GIPAM GmbH. GIPAM GmbH was contracted by Zentiva for the conduct of this study. BD is an employee of GWQ ServicePlus AG, which provided the anonymized research data used for this study. DZ has received compensation from Zentiva for consulting and Scientific Advisory Board participation. The authors declare no further competing interests.
Footnotes
Publisher’s note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Jabbour E, Kantarjian H (2025) Chronic myeloid leukemia: a review. JAMA 333(18):1618–1629. https://doi.oeg/10.1001/jama.2025.0220 [DOI] [PubMed]
- 2.Braun TP, Eide CA, Druker BJ (2020) Response and Resistance to BCR-ABL1-Targeted Therapies. Cancer Cell 37:530. 10.1016/J.CCELL.2020.03.006 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Bower H, Björkholm M, Dickman PW, Höglund M, Lambert PC, Andersson TM-L (2016) Life expectancy of patients with chronic myeloid leukemia approaches the life expectancy of the general population. J Clin Oncol 34:2851–2857 [DOI] [PubMed] [Google Scholar]
- 4.Vener C, Banzi R, Ambrogi F, Ferrero A, Saglio G, Pravettoni G et al (2020) First-line imatinib vs second- and third-generation TKIs for chronic-phase CML: a systematic review and meta-analysis. Blood Adv 4:2723. 10.1182/BLOODADVANCES.2019001329 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Levêque D, Becker G, Bilger K, Natarajan-Amé S (2020) Clinical pharmacokinetics and pharmacodynamics of dasatinib. Clin Pharmacokinet 59:849–856 [DOI] [PubMed] [Google Scholar]
- 6.Andersson P, Brisander M, Liljebris C, Jesson G, Lennernäs H (2024) Severe Impact of Omeprazole Timing on pH-Sensitive Dasatinib Absorption: Unveiling Substantial Drug–Drug Interaction. J Clin Pharmacol 65. 10.1002/JCPH.6173 [DOI] [PMC free article] [PubMed]
- 7.Larfors G, Andersson P, Jesson G, Liljebris C, Brisander M, Lennernäs H et al (2023) Despite warnings, co-medication with proton pump inhibitors and dasatinib is common in chronic myeloid leukemia, but XS004, a novel oral dasatinib formulation, provides reduced pH-dependence, minimizing undesirable drug–drug interactions. Eur J Haematol 111:644–654. 10.1111/EJH.14059 [DOI] [PubMed] [Google Scholar]
- 8.Sharma M, Holmes HM, Mehta HB, Chen H, Aparasu RR, Shih YT et al (2019) The concomitant use of tyrosine kinase inhibitors and proton pump inhibitors: prevalence, predictors, and impact on survival and discontinuation of therapy in older adults with cancer. Cancer 125:1155–1162 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Pajiep M, Conte C, Huguet F, Gauthier M, Despas F, Lapeyre-Mestre M (2021) Patterns of tyrosine kinase inhibitor utilization in newly treated patients with chronic myeloid leukemia: an exhaustive population-based study in France. Front Oncol 11:675609 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Lapka M, Khryakova A, Nazmutdinova A, Charalambous C, Bosak J, Sliva J Prescription Behaviour and Drug Interactions of Acid Suppressants With Tyrosine Kinase Inhibitors in Patients With Leukaemia: Data From Germany. EJHaem 2026;7:e70279. 10.1002/jha2.70279 [DOI] [PMC free article] [PubMed]
- 11.Dahlén T, Larfors G, Lennernäs H, Liljebris C, Von Euler M, Brisander M et al (2024) Frequency of comedication of proton pump inhibitors with crystalline dasatinib in chronic myeloid leukemia and effects on TKI-bioavailability. J Clin Oncol 42:6561–6561 10.1200/JCO.2024.42.16_suppl.6561
- 12.Apperley JF, Milojkovic D, Cross NCP, Hjorth-Hansen H, Hochhaus A, Kantarjian H et al (2025) 2025 European LeukemiaNet recommendations for the management of chronic myeloid leukemia: CHRONIC MYELOID LEUKEMIA. Leukemia. ;39:1797–813 [DOI] [PMC free article] [PubMed]
- 13.Cheng F, Xu Q, Li Q, Cui Z, Li W, Zeng F (2023) Adverse reactions after treatment with dasatinib in chronic myeloid leukemia: Characteristics, potential mechanisms, and clinical management strategies. Front Oncol 13:1113462 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Quan H, Sundararajan V, Halfon P, Fong A, Burnand B, Luthi J-C et al (2005) Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care 43:1130–1139 [DOI] [PubMed] [Google Scholar]
- 15.Chae J-W, Song CS, Kim H, Lee K-B, Seo B-S, Kim D-I (2011) Prediction of mortality in patients undergoing maintenance hemodialysis by Charlson Comorbidity Index using ICD-10 database. Nephron Clin Pract 117:379–384 [DOI] [PubMed] [Google Scholar]
- 16.Busque L, Beaudet M-É, Harnois M, Moussa H, Szuber N, Mollica L et al (2025) Switching TKIs during CML therapy is frequent, mostly driven by intolerance, and does not affect survival: a prospective Quebec registry study. Blood Cancer J 15:35. 10.1038/s41408-025-01242-8 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Brave M, Goodman V, Kaminskas E, Farrell A, Timmer W, Pope S et al (2008) Sprycel for chronic myeloid leukemia and Philadelphia chromosome–positive acute lymphoblastic leukemia resistant to or intolerant of imatinib mesylate. Clin Cancer Res 14:352–359 [DOI] [PubMed] [Google Scholar]
- 18.Haddad FG, Nasnas C, Sasaki K, Paul S, Issa GC, Rausch C et al (2024) Using proton pump inhibitors is not associated with adverse outcomes in patients with chronic myeloid leukemia treated with dasatinib. Clin Lymphoma Myeloma Leuk 24:484–487 [DOI] [PubMed] [Google Scholar]
- 19.Yoshimichi K, Taniguchi J, Naoko Y, Hiroki N, Tomoaki E, Kotaro N et al (2020) Influence of proton pump inhibitors and H2-receptor antagonists on the efficacy and safety of dasatinib in chronic myeloid leukemia patients. Int J Hematol 111:826–832 [DOI] [PubMed] [Google Scholar]
- 20.Saußele S, Kohlbrenner K, Vogelmann T, Schubert T, Incidence (2022) Prevalence, and Real-World Treatment Patterns in Chronic Myeloid Leukemia: Results from a Population-Representative German Claims Data Analysis. Oncol Res Treat 45:400–407. 10.1159/000524284 [DOI] [PubMed] [Google Scholar]
- 21.Pfirrmann M, Clark RE, Prejzner W, Lauseker M, Baccarani M, Saussele S et al (2020) The EUTOS long-term survival (ELTS) score is superior to the Sokal score for predicting survival in chronic myeloid leukemia. Leukemia 34:2138–2149 [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The datasets generated and/or analyzed during the current study are not publicly available since the findings of this study are extracted from health insurance claims. Data were available for research purposes from the sickness fund upon request, in an anonymized form. Due to restrictions around revealing patients’ confidential information, data were used under license for the current study, and so are neither publicly available nor can be shared further.