Meta-analysisindependent cross-checkfull text checked I-INS-06 in-depth review
Pooled studies of glargine and cancer
Insulin glargine and cancer risk in patients with diabetes: a meta-analysis.
Overview
Tang X et al.. PLoS One 2012 · PMID 23284776 · DOI 10.1371/journal.pone.0051814
Glargine was not linked to more cancer, including breast cancer.
Does the study support the specific claim? No · Methodological quality: low
Plain-language summary
Study explained simply
- What did the researchers want to find out?
Insulin glargine and cancer incidence
- How was the study done?
Meta-analysis of observational studies and RCTs with GRADE
- How many people, animals or samples were studied?
11 studies; 448,928 subjects; Diabetes patients
- What was actually found?
OR 0.81 any cancer.
- How large was the observed effect?
OR 0.81 (0.68-0.98)
- What do the results mean in everyday life?
No cancer increase.
- What are the study's limitations?
Mostly short-term observational data; very high heterogeneity; no study quality assessment; dose and long-term use not analysed; very low certainty.
- What does it say about Elawady's specific claim?
Contradicts C-INS-06.
- Do the results apply to Germany and Europe?
Includes European cohorts.
- What can explicitly not be concluded from this study?
That glargine protects against cancer, or anything about dose-dependent or long-term (>5 years) breast cancer risk.
Scientific in-depth review
1. Full reference
- Authors
- Tang X et al.
- Journal
- PLoS One
- Year
- 2012
- Identifier
- PMID 23284776 DOI 10.1371/journal.pone.0051814
- Verification
- verified – PubMed efetch 2026-10-25 (this agent)
- Retracted
- no
- Corrections
- none
- Extent checked
- full text
- Conflicts of interest and funding
Authors declared no competing interests. No funding statement present in the retrieved full text.
2. Research question
Insulin glargine and cancer incidence
3. Study design and population
- Study design
- Meta-analysis – Systematic review and random-effects meta-analysis of 1 RCT and 10 observational studies (9 cohorts, 1 case-control), with GRADE evidence profiles
- Population
Diabetes patients
- Country
Data from the Netherlands, France, UK, USA, Sweden, Taiwan/China, Italy, Canada, Germany and Scotland
- Period
Databases searched from inception to 12 March 2012; included studies published 2009-2012; follow-up of included studies mostly ~1.3-4.3 years (Mannucci 6.33 years)
- Sample size
11 studies; 448,928 subjects; 19,128 cancer patients (overall-cancer analysis: 10 studies, 433,701 participants)
- Inclusion and exclusion criteria
Observational studies and RCTs in people with diabetes comparing insulin glargine vs non-glargine insulin (human insulin, other analogues, isophane/NPH), published as peer-reviewed full text, reporting overall or site-specific cancer incidence; no language restriction; latest report used for duplicate populations. Two reviewers screened and extracted independently.
4. Methods and endpoints
- Exposure / intervention
Glargine
- Comparator
Non-glargine insulin
- Endpoints
Cancer incidence
- Statistical methods
Random-effects pooled ORs (Stata 12); heterogeneity by chi-square (P<0.10) and I2; Begg and Egger tests planned; subgroup analyses by comparator; sensitivity analyses (observational only, cohort only, excluding studies with prior-cancer patients, excluding self-reported exposure); GRADE per outcome. Only 8 of 11 studies adjusted for confounders; whether crude or adjusted estimates were pooled is not stated per study.
5. Actual results
- Results
Overall cancer (10 studies): OR 0.81 (95% CI 0.68-0.98), I2 = 93%, very low-quality evidence. vs other analogues OR 0.76 (0.62-0.93); vs human insulin OR 0.64 (0.60-0.68); vs isophane OR 0.67 (0.43-1.07). Breast cancer (8 studies): OR 0.99 (0.68-1.46), I2 = 79.9%; observational only OR 1.02 (0.68-1.53); excluding studies with prior-cancer patients OR 0.77 (0.49-1.21). Prostate OR 0.94 (0.63-1.42); pancreatic OR 1.08 (0.80-1.44); respiratory OR 0.91 (0.59-1.41); colorectal OR 0.69 (0.56-0.85); gastrointestinal OR 0.70 (0.51-0.95); hepatobiliary OR 0.51 (0.37-0.70); bladder OR 0.60 (0.37-0.99). All outcomes graded very low quality.
- Absolute risks
GRADE table: overall cancer 15,979/363,228 (4.4%) non-glargine vs 2,903/70,473 (4.1%) glargine; risk difference 8 fewer per 1000 (1 to 14 fewer). Breast cancer 1,104/241,976 (0.46%) vs 260/42,426 (0.61%); 0 fewer per 1000 (1 fewer to 2 more).
- Authors' conclusion
Results do not support a link between glargine and increased cancer risk; confidence very low.
6. Strengths
- Five databases plus manual/Google Scholar search, no language restriction
- Duplicate independent screening and data extraction
- GRADE evidence profile with absolute effects per outcome
- Several pre-specified sensitivity and subgroup analyses
7. Weaknesses
- Very high heterogeneity (I2 = 93% overall cancer; ~80% breast cancer)
- No study-level risk-of-bias tool; 3 of 11 studies unadjusted for confounders; few adjusted for metformin or diabetes type
- Short follow-up (mostly 1.3-4 years) relative to cancer latency
- No dose-response or duration analysis despite one included study (Suissa) suggesting longer-term breast cancer risk
- Publication-bias tests announced but not reported
- Excluded-studies list not provided
- Lower odds for several cancers are implausibly protective and likely reflect confounding by indication/channeling and time-related biases
- Funding not reported
8. Risk of bias
- Orienting tool
- AMSTAR 2
- Type of assessment
- structured assessment based on the full text by a single reviewer (no formal double review)
- Full formal assessment
- no
- Overall risk of bias
- high
- Rationale
AMSTAR 2 overall confidence: critically low (critical flaws in items 7, 9 and 11). Overall raised from some_concerns (abstract-based) to high. Single-reviewer structured judgement from the full text; a formal assessment requires two independent reviewers and the protocol.
| Domain | Judgement | Rationale |
|---|---|---|
| 1 PICO components | low | Population, glargine exposure, non-glargine comparator and cancer outcomes defined. |
| 2 Protocol established a priori (critical) | some concerns | States methods were 'specified in advance and protocol-defined', but no registered or published protocol is cited. |
| 3 Explanation of included designs | some concerns | Both RCTs and observational studies included without explicit justification. |
| 4 Comprehensive search (critical) | low | Five databases, reference lists, Google Scholar, no language restriction; restricted to peer-reviewed full texts. |
| 5 Duplicate selection | low | Two independent investigators with third-reviewer arbitration. |
| 6 Duplicate extraction | low | Two investigators extracted independently. |
| 7 List of excluded studies with reasons (critical) | high | Only a flow diagram; no list of excluded full texts with reasons. |
| 8 Description of included studies | low | Table 1 gives country, design, exposure source, n, comparator, age, follow-up and covariates. |
| 9 Risk of bias of individual studies (critical) | high | No study-level RoB tool (e.g. NOS/ROBINS); RoB only considered within GRADE downgrading. |
| 10 Funding of included studies | high | Not reported. |
| 11 Appropriate meta-analytic methods (critical) | high | Random effects used, but RCT and observational data pooled together, adjusted and unadjusted estimates mixed, and pooling performed despite I2 = 93%. |
| 12 Impact of RoB on results | some concerns | Sensitivity analyses on design and prior cancer, but not on adjustment status or study quality. |
| 13 RoB considered in interpretation (critical) | some concerns | Very low certainty acknowledged, yet discussion reports 'reduced odds' of cancer without stressing likely confounding. |
| 14 Heterogeneity explained | some concerns | Heterogeneity acknowledged; partly explored by comparator subgroup; largely unexplained. |
| 15 Publication bias (critical) | some concerns | Begg/Egger tests stated in methods but results not reported; authors note publication bias cannot be excluded. |
| 16 Conflicts of interest of review | some concerns | No competing interests declared; funding source not stated in retrieved text. |
Full-text review (25 October 2026): Added full search/selection methods, per-outcome ORs incl. site-specific cancers and comparator subgroups, I2 values, GRADE absolute effects, COI statement; AMSTAR 2 itemised (critically low; bias raised to high); noted absence of dose-response and long-term analysis.
9. Assessment of the statistical interpretation
The upper CI bound for overall cancer (0.98) excludes a large increase, but heterogeneity is extreme and inputs are mostly short-term observational data with variable adjustment, so the 'protective' ORs should not be read causally. The breast-cancer CI (0.68-1.46) is compatible with neither benefit nor a moderate harm. The study does not test dose-dependence.
10. Relevance to the specific claim
Contradicts C-INS-06.
11. Independent contrary findings
12. Transferability to Germany and Europe
Most data from European prescription databases (NL, FR, UK, SE, DE, IT, Scotland); directly transferable to EU insulin users, with the stated limitations.
13. Overall methodological judgement
Low-quality meta-analysis (critically low AMSTAR 2 confidence) whose consistent finding is the absence of a clear overall cancer increase with glargine over short follow-up; it neither tests dose-response nor resolves the breast cancer question.
| Multidimensional assessment | |
|---|---|
| Authenticity of the source | verified |
| Methodological quality | low |
| Risk of bias | high |
| Precision of results | moderate |
| Directness | direct |
| Reproducibility | partly replicated |
| Consistency with the overall evidence | inconsistent |
| Regional transferability | high |
| Adequacy of the interpretation (by the archive) | not applicable |
14. Open questions
Long-term (>5 years) and dose-dependent breast cancer risk; residual confounding by indication; effect of diabetes type and metformin co-treatment.
15. Linked fact checks
- Insulin glargine increases cancer risk in a dose-dependent way (and long-term use increases breast cancer). – Studies contradicting or limiting the claim