Dense tissue around pancreatic tumours can block chemotherapy in mice

Overview

Olive KP et al.. Science 2009 · PMID 19460966 · DOI 10.1126/science.1171362

Reducing the tumour's supporting tissue briefly let more drug reach the tumour.

Does the study support the specific claim? Partly · Methodological quality: moderate

Original publication

Plain-language summary

Scientific in-depth review

1. Full reference

Authors

Olive KP et al.

Journal

Science

Year

2009

Identifier

PMID 19460966 DOI 10.1126/science.1171362

Verification

verified – PubMed esummary + efetch 2026-10-25

Retracted

no

Corrections

none

Extent checked

full text

Conflicts of interest and funding

Funding: Lustgarten Foundation, University of Cambridge, Cancer Research UK, Li Ka Shing Foundation and Hutchison Whampoa, NIH grants, Mead Foundation; preliminary studies by UPenn/GSK ADDI grant. No competing-interest statement in the available text (IPI-926 is an industry compound; manufacturer relationship not stated here).

2. Research question

Does tumour stroma limit chemotherapy delivery in pancreatic cancer?

3. Study design and population

Study design

Animal study – Preclinical animal study: genetically engineered KPC mouse model (Kras/p53) vs transplantation models; perfusion imaging, pharmacokinetics, histology, and intervention survival study; plus analysis of human PDA specimens

Population

KPC mice with autochthonous pancreatic ductal adenocarcinoma; mice with transplanted (xenograft, orthotopic, syngeneic) tumours; 18 human PDA specimens vs normal pancreas (5) and chronic pancreatitis (5)

Country

UK/USA/Germany (animal work at Cancer Research UK Cambridge Research Institute, UPenn, FHCRC, University Hospital Dresden)

Period

not stated

Sample size

Small groups, e.g. 17 KPC mice on gemcitabine; perfusion studies N=4-8 per model; survival study IPI-926/gemcitabine n=17, gemcitabine n=10, IPI-926 n=10; 18 human PDA specimens

Inclusion and exclusion criteria

not reported in main text (methods in online supplement, not in the file)

4. Methods and endpoints

Exposure / intervention

Hedgehog/Smoothened inhibitor IPI-926 (40 mg/kg daily oral) plus gemcitabine

Comparator

Vehicle, gemcitabine alone, IPI-926 alone; transplanted tumour models

Endpoints

Tumour growth (ultrasound), intratumoral gemcitabine metabolites (HPLC, 19F NMR), perfusion (lectin, doxorubicin, contrast ultrasound, DCE-MRI), microvessel density, apoptosis, survival, liver metastases

Statistical methods

Mann-Whitney U, log-rank test, Fisher's exact test; no sample-size calculation, randomisation or blinding reported in main text

5. Actual results

Results

Gemcitabine inhibited all transplanted tumours but 15/17 KPC tumours grew like controls; active metabolite dFdCTP undetectable in KPC tumours. Only 32% of CD31+ vessels in KPC tumours were perfused vs 78% (transplanted) and 100% (normal pancreas). IPI-926 depleted desmoplastic stroma, increased microvessel density and raised intratumoral gemcitabine metabolites by 60% (P=.04). Combination extended median survival from 11 to 25 days (P=.001; HR 0.157), 14/17 tumours transiently shrank (vs 2/10 with each drug alone), fewer liver metastases (P=.015). Vascular density returned to control levels at endpoint and most tumours resumed growth. Human PDA specimens (n=18) were hypovascular vs normal pancreas and chronic pancreatitis.

Absolute risks

not applicable (animal study); median survival 11 vs 25 days

Authors' conclusion

Inefficient drug delivery due to poorly perfused, stroma-rich tumour architecture may be an important contributor to chemoresistance in pancreatic cancer; stromal depletion transiently improved delivery and efficacy of gemcitabine.

6. Strengths

• Autochthonous model resembling human PDA
• Several independent perfusion methods converge
• Direct measurement of intratumoral drug metabolites
• Survival endpoint

7. Weaknesses

• Mouse model; one tumour type (PDA)
• Small groups; randomisation, blinding and sample-size planning not reported
• Effect transient; tumours adapted (vascular density returned to baseline)
• Methods only in supplement (not available in file); available text is an author manuscript with unresolved editorial queries ('of what???')
• Stroma studied is tumour-induced desmoplasia, not pre-existing tissue fibrosis
• COI statement not available

8. Risk of bias

Orienting tool

none applicable (preclinical animal study; structured judgement, SYRCLE-oriented)

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

some concerns

Rationale

Internal validity has some concerns (unreported randomisation/blinding, small n); indirectness to human clinical outcomes is very high. Changed from 'not_assessable' (abstract) to 'some_concerns'. Single-reviewer structured judgement from the full text; a complete formal assessment requires two independent reviewers and the review/study protocol.

Full-text review (25 October 2026): File is the NIH author manuscript (correct article, contains editorial placeholders; supplementary methods absent). Added model details, group sizes, perfusion data (32% vs 78%/100%), +60% gemcitabine metabolites, survival 11 vs 25 days (HR 0.157), transient effect and vascular rebound, human specimen n=18, funding. Bias changed not_assessable -> some_concerns (structured preclinical judgement); precision set low.

9. Assessment of the statistical interpretation

Hypothesis-generating preclinical evidence with small groups; survival gain was a few days in mice and transient. Shows a stromal drug-delivery barrier in one tumour type, not a general mechanism of chemoresistance.

10. Relevance to the specific claim

Partial support for the idea that dense (fibrotic) tumour stroma can hinder chemotherapy, in specific tumours.

11. Independent contrary findings

Many cancers are highly chemosensitive (I-CHM-05; A-CHM-14).

12. Transferability to Germany and Europe

Not applicable (animal model); human relevance limited.

13. Overall methodological judgement

Shows that tumour-induced desmoplastic stroma can impede chemotherapy delivery in pancreatic cancer in mice; it does not show that fibrosis causes cancer or that this explains chemoresistance generally.

14. Open questions

Whether stromal targeting improves outcomes in patients (not assessed in this paper); applicability to other tumour types.

15. Linked fact checks

• Cancer cells resist chemotherapy because of their fibrotic interstitial (stromal) environment; fibrosis strangles cells, which then turn cancerous and resist chemo and radiation. – Studies with mixed or unclear results