The use of deep learning models to predict progression-free survival in patients with neuroendocrine tumors

Abstract

Aim: The RAISE project assessed whether deep learning could improve early progression-free survival (PFS) prediction in patients with neuroendocrine tumors. Patients & methods: Deep learning models extracted features from CT scans from patients in CLARINET (NCT00353496) (n = 138/204). A Cox model assessed PFS prediction when combining deep learning with the sum of longest diameter ratio (SLDr) and logarithmically transformed CgA concentration (logCgA), versus SLDr and logCgA alone. Results: Deep learning models extracted features other than lesion shape to predict PFS at week 72. No increase in performance was achieved with deep learning versus SLDr and logCgA models alone. Conclusion: Deep learning models extracted relevant features to predict PFS, but did not improve early prediction based on SLDr and logCgA.

Plain language summary

The use of deep learning models to predict progression-free survival in patients with neuroendocrine tumors: Neuroendocrine tumors (NET) are slow-growing cancers. How well cancers respond to treatment is usually measured using ‘Response Evaluation Criteria in Solid Tumors (RECIST)’, which is based on measuring the size of tumors. RECIST is not well suited for assessing NETs as these tumors often grow slowly and rarely shrink significantly, so it is difficult to tell whether a treatment has any effect. A better way of measuring how well NETs are responding to treatment is needed, to ensure that patients receive the right treatment as early as possible.

The RAISE project aimed to use a type of artificial intelligence (AI) called ‘deep learning’ to examine images of NETs, taken from patients in a clinical trial of treatment with lanreotide, to help predict how they might respond to treatment. These images were analyzed by the deep learning AI to see if there are any features of tumors, other than shape or size, that may help to predict response to treatment.

The project showed that this technology can detect features in images of NETs, other than the shape and size of tumors, that are useful for predicting how well a treatment might work for an individual patient. However, this technology could not improve prediction of how well a treatment would work at an earlier stage compared with other currently used indicators.

Overall, further research and work is needed to improve this technology. However, these results show that deep learning may have the potential to improve prediction of treatment response in patients with NETs.

Keywords::

• artificial intelligence
• deep learning
• neuroendocrine tumors
• progression-free survival
• RECIST

Author contributions

Substantial contributions to study conception and design: M Pavel, C Dromain, M Ronot, N Schaefer, D Mandair, D Gueguen, D Elvira, S Jégou, F Balazard, O Dehaene & K Schutte; substantial contributions to analysis and interpretation of the data: M Pavel, C Dromain, M Ronot, N Schaefer, D Mandair, D Gueguen, D Elvira, S Jégou, F Balazard, O Dehaene & K Schutte; drafting the article or revising it critically for important intellectual content: M Pavel, C Dromain, M Ronot, N Schaefer, D Mandair, D Gueguen, D Elvira, S Jégou, F Balazard, O Dehaene & K Schutte; final approval of the version of the article to be published: M Pavel, C Dromain, M Ronot, N Schaefer, D Mandair, D Gueguen, D Elvira, S Jégou, F Balazard, O Dehaene & K Schutte.

Acknowledgements

The authors thank all patients involved in the study, as well as their caregivers, care team, investigators, and research staff in participating institutions. The authors also thank Julie Benzimra, Thibaut Emorine, Lise Minssen, and Ilan Obadia for the annotation of images in the study.

Financial & competing interests disclosure

This study was sponsored by Ipsen. The collaboration between Ipsen and Owkin was funded by Ipsen. M Pavel: participated in advisory boards for, and received honoraria from AAA, Amgen, Boehringer Ingelheim, Eli Lilly, Ipsen, Lexicon, Novartis, Pfizer, Riemser; C Dromain: received consultancy fees and honoraria from Ipsen; M Ronot: received honoraria from Alexion Pharmaceuticals, Canon-Toshiba, GE Healthcare, Guerbet, Ipsen, Servier, Sirtex; N Schaefer, D Mandair: none to declare; D Gueguen: Former employee of Ipsen; D Elvira: employee of Ipsen; S Jégou, O Dehaene: former employees of Owkin; F Balazard & K Schutte: employees of Owkin. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed.

The authors thank Izzy Hawkes of Costello Medical, UK, for medical writing support and editorial assistance, which was sponsored by Ipsen in accordance with Good Publication Practice guidelines.

Ethical conduct of research

Data are from the CLARINET study. The CLARINET study was approved by all relevant local ethical committees. Consent was obtained from each patient after full explanation of the purpose and nature of all procedures used.

Data sharing statement

The authors certify that this manuscript reports the secondary analysis of clinical trial data that have been shared with them, and that the use of this shared data is in accordance with the terms (if any) agreed upon their receipt. The source of this data is: CLARINET phase III trial (NCT00353496)

Qualified researchers may request access to patient-level study data that underlie the results reported in this publication. Additional relevant study documents, including the clinical study report, study protocol with any amendments, annotated case report form, statistical analysis plan and dataset specifications may also be made available. Patient level data will be anonymized, and study documents will be redacted to protect the privacy of study participants.

Where applicable, data from eligible studies are available 6 months after the studied medicine and indication have been approved in the US and EU or after the primary manuscript describing the results has been accepted for publication, whichever is later.

Further details on Ipsen’s sharing criteria, eligible studies and process for sharing are available here (https://vivli.org/members/ourmembers/).

Any requests should be submitted to www.vivli.org for assessment by an independent scientific review board.

Restrictions apply to the availability of these data since the data underlying this publication were provided by Owkin under contract to Ipsen.

Additional information

Funding

This study was sponsored by Ipsen. The collaboration between Ipsen and Owkin was funded by Ipsen. M Pavel: participated in advisory boards for, and received honoraria from AAA, Amgen, Boehringer Ingelheim, Eli Lilly, Ipsen, Lexicon, Novartis, Pfizer, Riemser; C Dromain: received consultancy fees and honoraria from Ipsen; M Ronot: received honoraria from Alexion Pharmaceuticals, Canon-Toshiba, GE Healthcare, Guerbet, Ipsen, Servier, Sirtex; N Schaefer, D Mandair: none to declare; D Gueguen: Former employee of Ipsen; D Elvira: employee of Ipsen; S Jégou, O Dehaene: former employees of Owkin; F Balazard & K Schutte: employees of Owkin. The authors have no other relevant affiliations or financial involvement with any organization or entity with a financial interest in or financial conflict with the subject matter or materials discussed in the manuscript apart from those disclosed