MRI radiomics in the prediction of therapeutic response to neoadjuvant therapy for locoregionally advanced rectal cancer: a systematic review

ABSTRACT

Introduction: The standard of care for locoregionally advanced rectal cancer is neoadjuvant therapy (NA CRT) prior to surgery, of which 10–30% experience a complete pathologic response (pCR). There has been interest in using imaging features, also known as radiomics features, to predict pCR and potentially avoid surgery. This systematic review aims to describe the spectrum of MRI studies examining high-performing radiomic features that predict NA CRT response.

Areas covered: This article reviews the use of pre-therapy MRI in predicting NA CRT response for patients with locoregionally advanced rectal cancer (T3/T4 and/or N1+). The primary outcome was to identify MRI radiomic studies; secondary outcomes included the power and the frequency of use of radiomic features.

Expert opinion: Advanced models incorporating multiple radiomics categories appear to be the most promising. However, there is a need for standardization across studies with regards to; the definition of NA CRT response, imaging protocols, and radiomics features incorporated. Further studies are needed to validate current radiomics models and to fully ascertain the value of MRI radiomics in the response prediction for locoregionally advanced rectal cancer.

KEYWORDS:

• Radiomics
• rectal Cancer
• therapeutic response
• long course neoadjuvant chemoradiotherapy
• magnetic resonance imaging

Article highlights

• The standard of care for locoregionally advanced rectal cancer is neoadjuvant therapy prior to surgery

• Approximately 10-30% of patients who undergo neoadjuvant long course therapy (NA CRT) experience a complete pathologic response (pCR)

• There has been interest in using imaging features, also known as radiomics features, to predict pCR to NA CRT and therefore potentially avoid surgery

• 1^st order features that distinguished between response groups included kurtosis, uniformity, energy, skewness and entropy.

• Second and higher order features were of prognostic significance in most studies.

• Combining radiomics features into a ‘radiomics signature’ further enhanced response prediction (AUCs 0.72-0.91).

• There is a need for standardisation across studies in regard to the definition of NA CRT response, imaging protocols, and radiomics features incorporated

Declaration of interest

The authors have no 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. This includes employment, consultancies, honoraria, stock ownership or options, expert testimony, grants or patents received or pending, or royalties.

Reviewer disclosures

Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.