CoChat
Sign Up
PreprintCited by 12AI in cancer detectionRadiomics and Machine Learning in Medical ImagingPARP inhibition in cancer therapy

Direct prediction of Homologous Recombination Deficiency from routine histology in ten different tumor types with attention-based Multiple Instance Learning: a development and validation study

Open access
Lavinia Loeffler, Chiara Maria, El Nahhas, Omar S.M., Muti, Hannah Sophie, Seibel, Tobias, Cifci, Didem, Treeck, Marko van, Gustav, Marco, Carrero, Zunamys I. +7 more·2023·medRxiv
DOI 10.1101/2023.03.08.23286975PMC PMC10029072
Open this paper in CoChat — free accountOpen PDF

Abstract

Background: Homologous Recombination Deficiency (HRD) is a pan-cancer predictive biomarker that identifies patients who benefit from therapy with PARP inhibitors (PARPi). However, testing for HRD is highly complex. Here, we investigated whether Deep Learning can predict HRD status solely based on routine Hematoxylin & Eosin (H&E) histology images in ten cancer types. Methods: We developed a fully automated deep learning pipeline with attention-weighted multiple instance learning (attMIL) to predict HRD status from histology images. A combined genomic scar HRD score, which integrated loss of heterozygosity (LOH), telomeric allelic imbalance (TAI) and large-scale state transitions (LST) was calculated from whole genome sequencing data for n=4,565 patients from two independent cohorts. The primary statistical endpoint was the Area Under the Receiver Operating Characteristic curve (AUROC) for the prediction of genomic scar HRD with a clinically used cutoff value. Results: We found that HRD status is predictable in tumors of the endometrium, pancreas and lung, reaching cross-validated AUROCs of 0.79, 0.58 and 0.66. Predictions generalized well to an external cohort with AUROCs of 0.93, 0.81 and 0.73 respectively. Additionally, an HRD classifier trained on breast cancer yielded an AUROC of 0.78 in internal validation and was able to predict HRD in endometrial, prostate and pancreatic cancer with AUROCs of 0.87, 0.84 and 0.67 indicating a shared HRD-like phenotype is across tumor entities. Conclusion: In this study, we show that HRD is directly predictable from H&E slides using attMIL within and across ten different tumor types.

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Integrity signals

Registry

Crossref · 10.1101/2023.03.08.23286975

Cites

3 of 3 references verified0 citation errors · 0 not checkable

Statistics

Claims the paper makes

The claims and statistics of this paper haven’t been checked yet.

Run the check — free, no account needed. It reads the open-access text and checks the sources it cites.

A few free checks a day · papers already verified open instantly · results are public

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • The choice of tessellation of whole slide images at 256 μm × 256 μm resolution was motivated by recent successful applications of foundation models on H&E slides in predicting colorectal cancer MSI and pan-cancer homologous recombination deficiency.

    “Our choice of tessellation of whole slide images at this resolution was motivated by recent successful applications of foundation models on H&E slides, as demonstrated by studies predicting colorectal cancer MSI69 and pan-cancer homologous recombination deficiency70.”

    From Predicting the tumor microenvironment composition and immunotherapy response in non-small cell lung cancer from digital histopathology images (2024) · cites this paper as [70]

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

The self-check needs the paper’s full text.

Cites

Flagged first, then by citations

Research works printed but not linked: 49

  1. [1]Frey MK, Pothuri B. Homologous recombination deficiency (HRD) testing in ovarian cancer clinical practice: a review of the literature. Gynecol Oncol Res Pract. 2017 Feb 22; 4: 4. 28250960 PMC5322589
  2. [2]Hoeijmakers JH. Genome maintenance mechanisms for preventing cancer. Nature. 2001 May 17; 411(6835): 366– 74. 11357144
  3. [3]Rose M, Burgess JT, O’Byrne K, Richard DJ, Bolderson E. PARP Inhibitors: Clinical Relevance, Mechanisms of Action and Tumor Resistance. Front Cell Dev Biol. 2020 Sep 9; 8: 564601. 33015058 PMC7509090
  4. [4]Dedes KJ, Wilkerson PM, Wetterskog D, Weigelt B, Ashworth A, Reis-Filho JS. Synthetic lethality of PARP inhibition in cancers lacking BRCA1 and BRCA2 mutations. Cell Cycle. 2011 Apr 15; 10(8): 1192– 9. 21487248 PMC3117132
  5. [5]Leary A, Auguste A, Mesnage S. DNA damage response as a therapeutic target in gynecological cancers. Curr Opin Oncol. 2016 Sep; 28(5): 404– 11. 27455135
  6. [6]Park W, Chen J, Chou JF, Varghese AM, Yu KH, Wong W, Genomic Methods Identify Homologous Recombination Deficiency in Pancreas Adenocarcinoma and Optimize Treatment Selection. Clin Cancer Res. 2020 Jul 1; 26(13): 3239– 47. 32444418 PMC7380542
  7. [7]Takaya H, Nakai H, Takamatsu S, Mandai M, Matsumura N. Homologous recombination deficiency status-based classification of high-grade serous ovarian carcinoma. Sci Rep. 2020 Feb 17; 10(1): 2757. 32066851 PMC7026096
  8. [8]Tutt ANJ, Garber JE, Kaufman B, Viale G, Fumagalli D, Rastogi P, Adjuvant Olaparib for Patients with BRCA1- or BRCA2-Mutated Breast Cancer. N Engl J Med. 2021 Jun 24; 384(25): 2394– 405. 34081848 PMC9126186
  9. [9]Ledermann JA. PARP inhibitors in ovarian cancer. Ann Oncol. 2016 Apr; 27 Suppl 1: i40– 4. 27141070
  10. [10]Stewart MD, Merino Vega D, Arend RC, Baden JF, Barbash O, Beaubier N, Homologous Recombination Deficiency: Concepts, Definitions, and Assays. Oncologist. 2022 Mar 11; 27(3): 167– 74. 35274707 PMC8914493
  11. [11]Miller RE, Leary A, Scott CL, Serra V, Lord CJ, Bowtell D, ESMO recommendations on predictive biomarker testing for homologous recombination deficiency and PARP inhibitor benefit in ovarian cancer. Ann Oncol. 2020 Dec; 31(12): 1606– 22. 33004253
  12. [12]Wagener-Ryczek S, Merkelbach-Bruse S, Siemanowski J. Biomarkers for Homologous Recombination Deficiency in Cancer. J Pers Med [Internet]. 2021 Jun 28; 11(7). Available from: 10.3390/jpm11070612 PMC8304859 34203281 · DOI
  13. [13]Fuh K, Mullen M, Blachut B, Stover E, Konstantinopoulos P, Liu J, Homologous recombination deficiency real-time clinical assays, ready or not? Gynecol Oncol. 2020 Dec; 159(3): 877– 86. 32967790
  14. [14]Alexandrov LB, Kim J, Haradhvala NJ, Huang MN, Tian Ng AW, Wu Y, The repertoire of mutational signatures in human cancer. Nature. 2020 Feb; 578(7793): 94– 101. 32025018 PMC7054213
  15. [15]Gulhan DC, Lee JJK, Melloni GEM, Cortés-Ciriano I, Park PJ. Detecting the mutational signature of homologous recombination deficiency in clinical samples. Nat Genet. 2019 May; 51(5): 912– 9. 30988514
  16. [16]Abkevich V, Timms KM, Hennessy BT, Potter J, Carey MS, Meyer LA, Patterns of genomic loss of heterozygosity predict homologous recombination repair defects in epithelial ovarian cancer. Br J Cancer. 2012 Nov 6; 107(10): 1776– 82. 23047548 PMC3493866
  17. [17]Birkbak NJ, Wang ZC, Kim JY, Eklund AC, Li Q, Tian R, Telomeric allelic imbalance indicates defective DNA repair and sensitivity to DNA-damaging agents. Cancer Discov. 2012 Apr; 2(4): 366– 75. 22576213 PMC3806629
  18. [18]Popova T, Manié E, Rieunier G, Caux-Moncoutier V, Tirapo C, Dubois T, Ploidy and large-scale genomic instability consistently identify basal-like breast carcinomas with BRCA1/2 inactivation. Cancer Res. 2012 Nov 1; 72(21): 5454– 62. 22933060
  19. [19]Westphalen CB, Fine AD, André F, Ganesan S, Heinemann V, Rouleau E, Pan-cancer Analysis of Homologous Recombination Repair-associated Gene Alterations and Genome-wide Loss-of-Heterozygosity Score. Clin Cancer Res. 2022 Apr 1; 28(7): 1412– 21. 34740923 PMC8982267
  20. [20]Sztupinszki Z, Diossy M, Krzystanek M, Reiniger L, Csabai I, Favero F, Migrating the SNP array-based homologous recombination deficiency measures to next generation sequencing data of breast cancer. NPJ Breast Cancer. 2018 Jul 2; 4: 16. 29978035 PMC6028448
  21. [21]Zhao EY, Shen Y, Pleasance E, Kasaian K, Leelakumari S, Jones M, Homologous Recombination Deficiency and Platinum-Based Therapy Outcomes in Advanced Breast Cancer. Clin Cancer Res. 2017 Dec 15; 23(24): 7521– 30. 29246904
  22. [22]Nguyen L, W M Martens J, Van Hoeck A, Cuppen E. Pan-cancer landscape of homologous recombination deficiency. Nat Commun. 2020 Nov 4; 11(1): 5584. 33149131 PMC7643118
  23. [23]Pellegrino B, Herencia-Ropero A, Llop-Guevara A, Pedretti F, Moles-Fernandez A, Viaplana C, Preclinical In Vivo Validation of the RAD51 Test for Identification of Homologous Recombination-Deficient Tumors and Patient Stratification. Cancer Res. 2022 Apr 15; 82(8): 1646– 57. 35425960 PMC7612637
  24. [24]Graeser M, McCarthy A, Lord CJ, Savage K, Hills M, Salter J, A marker of homologous recombination predicts pathologic complete response to neoadjuvant chemotherapy in primary breast cancer. Clin Cancer Res. 2010 Dec 15; 16(24): 6159– 68. 20802015 PMC3432445
  25. [25]How JA, Jazaeri AA, Fellman B, Daniels MS, Penn S, Solimeno C, Modification of Homologous Recombination Deficiency Score Threshold and Association with Long-Term Survival in Epithelial Ovarian Cancer. Cancers [Internet]. 2021 Feb 24; 13(5). Available from: 10.3390/cancers13050946 PMC7956737 33668244 · DOI
  26. [26]Schmauch B, Romagnoni A, Pronier E, Saillard C, Maillé P, Calderaro J, A deep learning model to predict RNA-Seq expression of tumours from whole slide images. Nat Commun. 2020 Aug 3; 11(1): 3877. 32747659 PMC7400514
  27. [27]Loeffler CML, Ortiz Bruechle N, Jung M, Seillier L, Rose M, Laleh NG, Artificial Intelligence-based Detection of FGFR3 Mutational Status Directly from Routine Histology in Bladder Cancer: A Possible Preselection for Molecular Testing? Eur Urol Focus. 2022 Mar; 8(2): 472– 9. 33895087
  28. [29]Fu Y, Jung AW, Torne RV, Gonzalez S, Vöhringer H, Shmatko A, Pan-cancer computational histopathology reveals mutations, tumor composition and prognosis. Nature Cancer. 2020 Jul 27; 1— 11. 35122049
  29. [30]Kather JN, Heij LR, Grabsch HI, Loeffler C, Echle A, Muti HS, Pan-cancer image-based detection of clinically actionable genetic alterations. Nature Cancer. 2020 Aug 1; 1(8): 789– 99. 33763651 PMC7610412
  30. [31]Kather JN, Pearson AT, Halama N, Jäger D, Krause J, Loosen SH, Deep learning can predict microsatellite instability directly from histology in gastrointestinal cancer. Nat Med. 2019 Jul; 25(7): 1054– 6. 31160815 PMC7423299
  31. [32]Muti HS, Heij LR, Keller G, Kohlruss M, Langer R, Dislich B, Development and validation of deep learning classifiers to detect Epstein-Barr virus and microsatellite instability status in gastric cancer: a retrospective multicentre cohort study [ Internet]. The Lancet Digital Health. 2021. Available from: 10.1016/s2589-7500(21)00133-3 PMC8460994 34417147 · DOI · Paper page
  32. [33]Saillard C, Dubois R, Tchita O, Loiseau N, Garcia T, Adriansen A, Blind validation of MSIntuit, an AI-based pre-screening tool for MSI detection from histology slides of colorectal cancer [ Internet]. bioRxiv. 2022. Available from: 10.1101/2022.11.17.22282460.abstract PMC10628260 37932267 · DOI
  33. [34]Kleppe A, Skrede OJ, De Raedt S, Hveem TS, Askautrud HA, Jacobsen JE, A clinical decision support system optimising adjuvant chemotherapy for colorectal cancers by integrating deep learning and pathological staging markers: a development and validation study. Lancet Oncol. 2022 Sep; 23(9): 1221– 32. 35964620
  34. [35]Valieris R, Amaro L, Osorio CAB de T, Bueno AP, Rosales Mitrowsky RA, Carraro DM, Deep Learning Predicts Underlying Features on Pathology Images with Therapeutic Relevance for Breast and Gastric Cancer. Cancers [Internet]. 2020 Dec 9; 12(12). Available from: 10.3390/cancers12123687 PMC7763049 33316873 · DOI
  35. [36]Lazard T, Bataillon G, Naylor P, Popova T, Bidard FC, Stoppa-Lyonnet D, Deep learning identifies morphological patterns of homologous recombination deficiency in luminal breast cancers from whole slide images. Cell Rep Med. 2022 Dec 20; 3(12): 100872. 36516847 PMC9798078
  36. [37]Gao J, Aksoy BA, Dogrusoz U, Dresdner G, Gross B, Sumer SO, Integrative analysis of complex cancer genomics and clinical profiles using the cBioPortal. Sci Signal. 2013 Apr 2; 6(269): 11. PMC4160307 23550210
  37. [38]Cerami E, Gao J, Dogrusoz U, Gross BE, Sumer SO, Aksoy BA, The cBio cancer genomics portal: an open platform for exploring multidimensional cancer genomics data. Cancer Discov. 2012 May; 2(5): 401– 4. 22588877 PMC3956037
  38. [39]Riaz N, Blecua P, Lim RS, Shen R, Higginson DS, Weinhold N, Pan-cancer analysis of bi-allelic alterations in homologous recombination DNA repair genes. Nat Commun. 2017 Oct 11; 8(1): 857. 29021619 PMC5636842
  39. [42]Leiby JS, Hao J, Kang GH, Park JW, Kim D. Attention-based multiple instance learning with self-supervision to predict microsatellite instability in colorectal cancer from histology whole-slide images. Conf Proc IEEE Eng Med Biol Soc. 2022 Jul; 2022: 3068– 71. 36085965
  40. [43]Ilse M, Tomczak J, Welling M. Attention-based Deep Multiple Instance Learning. In: Dy J, Krause A, editors. Proceedings of the 35th International Conference on Machine Learning. PMLR; 10–15 Jul 2018. p. 2127– 36. (Proceedings of Machine Learning Research; vol. 80).
  41. [44]Rempel E, Kluck K, Beck S, Ourailidis I, Kazdal D, Neumann O, Pan-cancer analysis of genomic scar patterns caused by homologous repair deficiency (HRD). NPJ Precis Oncol. 2022 Jun 9; 6(1): 36. 35681079 PMC9184602
  42. [45]Kleppe A, Skrede OJ, De Raedt S, Liestøl K, Kerr DJ, Danielsen HE. Designing deep learning studies in cancer diagnostics. Nat Rev Cancer. 2021 Mar; 21(3): 199– 211. 33514930
  43. [46]Ng CKY, Piscuoglio S, Geyer FC, Burke KA, Pareja F, Eberle CA, The Landscape of Somatic Genetic Alterations in Metaplastic Breast Carcinomas. Clin Cancer Res. 2017 Jul 15; 23(14): 3859– 70. 28153863 PMC5511565
  44. [47]Ngoi NYL, Tan DSP. The role of homologous recombination deficiency testing in ovarian cancer and its clinical implications: do we need it? ESMO Open. 2021 Jun; 6(3): 100144. 34015643 PMC8141874
  45. [48]Loeffler CML, Gaisa NT, Muti HS, van Treeck M. Predicting Mutational Status of Driver and Suppressor Genes Directly from Histopathology With Deep Learning: A Systematic Study Across 23 Solid Tumor …. Frontiers in [Internet]. 2021; Available from: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC8889144/ PMC8889144 35251119
  46. [49]Takamatsu S, Brown JB, Yamaguchi K, Hamanishi J, Yamanoi K, Takaya H, Utility of Homologous Recombination Deficiency Biomarkers Across Cancer Types. JCO Precis Oncol. 2022 May; 6: e2200085. 35613413 PMC9200383
  47. [50]Moukarzel LA, Ferrando L, Da Cruz Paula A, Brown DN, Geyer FC, Pareja F, The genetic landscape of metaplastic breast cancers and uterine carcinosarcomas. Mol Oncol. 2021 Apr; 15(4): 1024– 39. 33021035 PMC8024717
  48. [51]Na B, Yu X, Withers T, Gilleran J, Yao M, Foo TK, Therapeutic targeting of BRCA1 and TP53 mutant breast cancer through mutant p53 reactivation. NPJ Breast Cancer. 2019 Apr 15; 5: 14. 30993195 PMC6465291
  49. [52]Lai Z, Brosnan M, Sokol ES, Xie M, Dry JR, Harrington EA, Landscape of homologous recombination deficiencies in solid tumours: analyses of two independent genomic datasets. BMC Cancer. 2022 Jan 3; 22(1): 13. 34979999 PMC8722117

Cited by

Sorted by citations

Authors
Lavinia Loeffler, C. M., El Nahhas, O. S. M., Muti, H. S., Seibel, T., Cifci, D., Treeck, M. V., Gustav, M., Carrero, Z. I., Gaisa, N. T., Lehmann, K.-V., Leary, A., Selenica, P., Reis-Filho, J. S., Bruechle, N. O., & Kather, J. N.
Affiliations
  • University Hospital Carl Gustav Carus (DE)
  • Universitätsklinikum Aachen (DE)
  • Else Kröner Fresenius Center for Digital Health (DE)
  • Institut Gustave Roussy (FR)
  • Memorial Sloan Kettering Cancer Center (US)
  • University of Leeds (GB)
Corresponding author
Kather, Jakob Nikolas
ORCID
Venue
medRxiv
Publisher
openRxiv
Published
2023-03-10
DOI
10.1101/2023.03.08.23286975
PMC
PMC10029072
OpenAlex
W4323822316
Cited by
12
References
3
Citations by year
2025: 3 · 2024: 6 · 2023: 3
Type
posted content
Language
en
Open access
Licence
cc-by-nc-nd
Locations
  • medRxiv (acceptedVersion, cc-by-nc-nd)
  • PubMed (publishedVersion)
  • PubMed Central (submittedVersion, cc-by-nc-nd)
Funders
  • Breast Cancer Research Foundation (CA008748, P50 CA247749)
  • Susan G. Komen
  • National Institute for Health and Care Research (NIHR213331, NIHR, NIHR213331)
  • Department of Health and Social Care
  • Deutscher Akademischer Austauschdienst ((SECAI, 57616814), 57616814)
  • Bundesministerium für Bildung und Forschung (57616814, 01KD2104C)
  • Bundesministerium für Gesundheit (ZMVI1-2520DAT111, 2520DAT111, DEEP LIVER ZMVI1-2520DAT111)
  • Deutsche Krebshilfe (#70113864, 2520DAT111)
  • National Institutes of Health (CA008748, P50 CA247749 01, NCI P50 CA247749 01)
  • Leeds Biomedical Research Centre (NIHR213331)
  • National Cancer Institute (CA247749, CA008748, P50-CA247749-01, P50CA247749, P30CA008748)
Topics
  • Physical Sciences › Computer Science › Artificial Intelligence › AI in cancer detection
  • Health Sciences › Medicine › Radiology, Nuclear Medicine and Imaging › Radiomics and Machine Learning in Medical Imaging
  • Health Sciences › Medicine › Oncology › PARP inhibition in cancer therapy
Keywords
homologous recombination deficiency, PARP inhibitors, deep learning, loss of heterozygosity, telomeric allelic imbalance, large-scale state transitions, endometrial cancer, pancreatic cancer, lung cancer, breast cancer, prostate cancer

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review