Generic versus branded imatinib in frontline chronic myeloid leukemia: A systematic review and meta‐analysis

Chronic myeloid leukemia (CML) affects approximately 1 to 2 per 100,000 adults annually, and imatinib has transformed the disease into a chronic condition with near‐normal life expectancy. ¹ , ² Branded imatinib exceeded $100,000 per year in the United States prior to patent expiry, and more than 100 CML experts called these prices unsustainable. ³ The European LeukemiaNet 2020 recommendations endorsed generic imatinib as the cost‐effective initial treatment for chronic‐phase (CP) CML. ⁴ Individual studies comparing generic and branded imatinib have reported conflicting results. The only prior systematic review, by Erçalışkan et al., ⁵ narratively summarized 36 papers but performed no quantitative pooling. We conducted the first meta‐analysis pooling frontline clinical outcomes of generic versus branded imatinib in CML‐CP.

This systematic review followed PRISMA 2020 guidelines ⁶ and was registered with PROSPERO. PubMed, Embase, and Scopus were searched from inception through January 2025 (Supporting Information S5: Table 6). Comparative observational studies enrolling adults with CML‐CP who received generic imatinib with at least one frontline outcome compared with branded imatinib were included. Two reviewers independently extracted data and assessed quality using the Newcastle‐Ottawa Scale (NOS; 0–9; Supporting Information S5: Table 2). Random‐effects meta‐analyses used the Mantel–Haenszel method with DerSimonian–Laird estimation. Pooled odds ratios (ORs) with 95% confidence intervals (CIs) were calculated for binary outcomes; hazard ratios (HRs) were pooled for event‐free survival using the inverse‐variance method. Heterogeneity was assessed with I ² and Cochran Q. Pre‐specified subgroup analyses were by geographic region and NOS quality. Leave‐one‐out sensitivity analyses were performed for both primary outcomes. Meta‐regression examined publication year and NOS score as moderators. Publication bias was assessed with funnel plots and trim‐and‐fill. Certainty of evidence was assessed using GRADE (Supporting Information S5: Table 4). Switch response and detailed safety data from this systematic review will be reported separately.

A total of 696 records were identified; after removing 292 duplicates and screening 404 records, 30 full texts were assessed (Supporting Information S5: Table 5) and 21 studies were included (Supporting Information S1: Figure 1). The 21 studies7, 8, 9, 10, 11, 12, 13, 14, 15 encompassed 5349 patients across 13 countries, published between 2013 and 2025 (Supporting Information S5: Table 1). NOS scores ranged from 2 to 9 (median, 7); 13 studies (62%) were rated good quality. Twelve frontline comparative studies contributed to at least one pooled analysis. Of the 16 studies included in quantitative pooling, fewer than half contributed to any individual endpoint, and several analyses relied on 3–5 studies.

Complete cytogenetic response (CCyR) was reported in 7 studies (3020 patients; generic, 771; branded, 2249). Generic imatinib was not associated with a difference in CCyR (OR, 1.05; 95% CI, 0.84–1.31; I ² = 0%; Q P = 0.778) (Figure 1A). The prediction interval was 0.80–1.38.

Figure 1.

Forest plots comparing generic versus branded imatinib in frontline chronic myeloid leukemia (CML)‐chronic phase (CP). (A) Complete cytogenetic response (CCyR; k = 7; odds ratio [OR] 1.05; 95% CI, 0.84–1.31; I ² = 0%). (B) Major molecular response (MMR; k = 8; OR 0.82; 95% CI, 0.56–1.18; I ² = 53.2%). Random‐effects model, Mantel–Haenszel method.

Major molecular response (MMR; BCR‐ABL1 ≤ 0.1% IS) was reported in eight studies (2545 patients; generic, 780; branded, 1765). Generic imatinib was not associated with a difference in MMR (OR, 0.82; 95% CI, 0.56–1.18; I ² = 53.2%; Q P = 0.036) (Figure 1B). The prediction interval was 0.30–2.22.

Overall survival was reported in five studies. Tangkitchot et al. ¹⁵ was excluded from pooling (multivariate HR of 422, a numerical artifact from near‐complete separation in 25 generic patients). Four studies (1881 patients; generic, 756; branded, 1125) were pooled (Supporting Information S2: Figure 2). Generic imatinib was not associated with a difference in mortality (OR, 1.12; 95% CI, 0.79–1.58; I ² = 0%).

Treatment failure, defined per original study criteria as lack of optimal response, loss of response, progression to accelerated or blast phase, or death, was reported in three studies (932 patients). Generic imatinib was associated with higher treatment failure (OR, 1.52; 95% CI, 1.08–2.12; I ² = 1.3%) (Supporting Information S3: Figure 3), driven by Pagnano et al. ¹² and Iqbal et al. ¹⁰ Event‐free survival, defined as freedom from loss of complete cytogenetic or MMR, progression, or death, was reported as HRs in three studies (634 patients). Pooled HR was 1.56 (95% CI, 1.05–2.32). Excluding Pagnano et al., ¹² whose branded cohort was a historical control from 2010 to 2011 compared with prospective generic enrollment in 2015–2018, yielded HR 1.31 (95% CI, 0.71–2.40; I ² = 0%). Early molecular response (BCR‐ABL1 ≤ 10% at 3 months) was reported in five studies (502 patients), including Eskazan 2017 which specifically compared early molecular response between generic and branded formulations. The pooled OR was 0.66 (95% CI, 0.23–1.86; I ² = 81.4%). Excluding Kircali et al. ¹¹ (generic, 32.2%; branded, 76.5%) yielded OR 1.00 (95% CI, 0.57–1.75; I ² = 0%). All pooled results are summarized in Table 1.

Table 1.

Pooled meta‐analysis results: frontline outcomes.

Outcome	k	Generic N	Branded N	Random‐effects OR/HR (95% CI)	I 2 (%)	Prediction interval
CCyR	7	771	2249	OR 1.05 (0.84–1.31)	0.0	0.80–1.38
MMR	8	780	1765	OR 0.82 (0.56–1.18)	53.2	0.30–2.22
OS (mortality)	4	756	1125	OR 1.12 (0.79–1.58)	0.0	0.63–1.96
Treatment failure	3	491	441	OR 1.52 (1.08–2.12)	1.3	0.71–3.22
EFS	3	352	282	HR 1.56 (1.05–2.32)	‐	‐
EMR/optimal 3m	5	223	279	OR 0.66 (0.23–1.86)	81.4	0.02–17.5

Abbreviations: CCyR, complete cytogenetic response; CI, confidence interval; EFS, event‐free survival; EMR, early molecular response; HR, hazard ratio; MMR, major molecular response; OR, odds ratio; OS, overall survival.

Among frontline studies reporting adverse events, no consistent pattern of increased toxicity was observed with generic imatinib. Danthala et al. ⁷ (n = 1211) reported lower edema rates with generics (generic, 4.8%; branded, 12.1%) with no Grade 3–4 events in either arm. Iqbal et al. ¹⁰ reported comparable Grade 3–4 hematologic toxicity (generic, 26%; branded, 30%). Dou et al. ⁹ reported similar adverse event profiles between arms. Available frontline safety data are summarized in Supporting Information S5: Table 5.

Subgroup analysis by region showed no difference for CCyR (Q = 0.05; P = 0.817) but a significant regional difference for MMR (Q = 6.56; P = 0.038). Leave‐one‐out analysis for CCyR yielded ORs ranging from 0.97 to 1.11; for MMR, from 0.73 to 0.92. Restricting to good‐quality studies (NOS ≥ 7) yielded CCyR OR 1.17 (95% CI, 0.87–1.57; k = 5) and MMR OR 0.92 (95% CI, 0.56–1.50; k = 6). Funnel plots showed no asymmetry; trim‐and‐fill estimated two missing studies for CCyR (adjusted OR, 1.09; 95% CI, 0.88–1.34) and one for MMR (adjusted OR, 0.77; 95% CI, 0.54–1.10). Full sensitivity and meta‐regression results are shown in Supporting Information S5: Table 3.

Publication year was associated with MMR effect size on meta‐regression (coefficient, +0.129/year; P = 0.007; R ² = 90.0%). NOS score was also associated with MMR effect size (coefficient, +0.228; P = 0.028; R ² = 72.2%). Earlier studies reported larger differences between generic and branded imatinib; more recent, higher quality studies reported smaller or no differences. Plausible explanations include improvements in generic manufacturing processes and regulatory oversight, evolving molecular monitoring standards in participating countries, and a shift toward more rigorous study designs in later publications.

In this first meta‐analysis pooling frontline outcomes of generic versus branded imatinib, generic imatinib was not associated with differences in CCyR, MMR, or OS. The CCyR result was particularly stable: zero between‐study variance, a narrow prediction interval, and no shift on leave‐one‐out analysis. The treatment failure signal (OR, 1.52) was the only adverse finding. This estimate was based on only three studies, GRADE certainty was rated low due to the observational design and potential confounding, and the result was driven largely by Pagnano et al., ¹² which used a historical branded control from a different era. The EFS result became non‐significant after exclusion of this study. These findings should be considered hypothesis‐generating. These data support the ELN recommendation of generic imatinib as cost‐effective frontline therapy, ⁴ while reinforcing the value of molecular monitoring during the first treatment year. All included studies were observational, only 7 of 21 adjusted for confounders, and the number of studies per analysis was small (k = 3–8). Formal publication bias assessment was limited by the small number of studies per endpoint, as funnel plot asymmetry testing has low power below k = 10. The term “generic imatinib” encompasses products from multiple manufacturers across different countries with potentially variable bioequivalence and quality profiles, which could not be disentangled. None of the included studies reported treatment‐free remission outcomes or formal intolerance comparisons, which remain important evidence gaps. Future prospective studies with standardized endpoints, concurrent controls, and manufacturer‐level data are warranted.

AUTHOR CONTRIBUTIONS

Shankar Biswas: Conceptualization; funding acquisition; investigation; validation; writing—review and editing; visualization; project administration; formal analysis; software; methodology; supervision; data curation; resources; writing—original draft. Yashasvi Srivastava: Conceptualization; investigation; funding acquisition; writing—original draft. Nayanika Tummala: Methodology; validation; visualization; writing—review and editing. Ayman Hamadttu: Software; formal analysis; project administration; resources; supervision.

CONFLICT OF INTEREST STATEMENT

The authors declare no conflicts of interest.

ETHICS STATEMENT

Ethics approval and consent to participate: Not applicable. This study is a systematic review and meta‐analysis of previously published data; no individual patient data were accessed, and no ethical approval was required.

FUNDING

This research did not receive any specific grant from funding agencies in the public, commercial, or not‐for‐profit sectors.

Supporting information

Supporting Information.

Supporting Information.

Supporting Information.

Supporting Information.

Supporting Information.

DATA AVAILABILITY STATEMENT

The data that support the findings of this study are available from the corresponding author upon reasonable request.