Ultrasound and Photoacoustic Imaging for Enhanced Differential Diagnosis of Rectal Cancer

The investigators hypothesize that a co-registered photoacoustic ultrasound endorectal device can significantly reduce unnecessary surgeries in rectal cancer patients with complete clinical response while maintaining high sensitivity in identifying those with residual cancer. The proposed new imaging technique combined with deep-learning algorithms will provide a real-time imaging tool to assist surgeons in predicting the neoadjuvant treatment response of rectal cancer patients. Such a tool will offer sensitive vascular and %sO2 contrast for identifying complete responses, thereby assisting surgeons in significantly improving sensitivity and specificity in surgery recommendations. The study will 1) rigorously validate the technique in a group of locally advanced rectal cancer (LARC) patients who will undergo surgical resection or "wait and watch" surveillance of their rectum based on the standard of care (SOC), and 2) explore the technique's potential in assessing a sub-group of LARC patients through the course of neoadjuvant treatment and in post-treatment surveillance. Successful completion of the project will result in a new imaging tool that can significantly improve the SOC by reducing unnecessary surgery without compromising cancer-related outcomes and thereby lowering morbidity and health care cost.

In this trial, the investigators will recruit patients with rectal cancer who have completed neoadjuvant therapy and will undergo surgery or nonoperative management. Approximately 86 patients will be recruited. The first group of 30 patients (group #1) will be used to train/validate identification algorithm. The second group of 56 patients (group #2) will be used as the testing cohort to train the reader team comprised of 3 surgeons and 1 radiologist. Eligible patients will include those with rectal cancer who received neoadjuvant therapy and were deemed to either have incomplete or indeterminant or clinical response based on standard endoscopy and axial imaging with magnetic resonance. PAM/US will be performed before their surgery or at the time of patients' follow up imaging.

In the investigation of group #2 patients, the study surgeon will review each patient's record and decide on surgery (clinical non-responders, non-cCR) for two types of patients of 1) those deemed non-cCR based on SOC at the treatment completion; 2) those deemed cCR at the treatment completion and are under "watch and wait" surveillance .

After initial training using group #1 data, the study surgeon will grade the tumor bed based on MRI and other clinical information as non-cCR or cCR at the time of SOC imaging after treatment completion. PAM/US imaging will be performed at the same time as the patients' SOC imaging. The study surgeon will grade the tumor bed as either PAM/US non-cCR or PAM/US cCR after PAM/US imaging.

For patients with surgery, the pathological outcomes of the patients will be assessed based on the Modified Ryan Scheme for Tumor Regression Score. Score 0, No viable cancer cells (pCR); Score 1, Single cells or rare small groups of cancer cells (near pCR); Score 2, Residual cancer with evident tumor regression, but more than single cells or rare small groups of cancer cells; Score 3, Extensive residual cancer with no evident tumor regression. Scores 2 and 3 are considered as non-pCRs. For patients without surgery, cCR without of any concerns of recurrence at the end of 2 year follow-up window will be considered as responder. The radiologist reader will also independently grade each case based on MRI.

The sensitivity and specificity as well as the area under the receiver operation characteristic curve (AUC) of the reader team based on the SOC will be evaluated. The sensitivity and specificity and AUC of the reader team based on PAM/US data will be evaluated. If the PAM/US methodology is validated from this prospective study, surgeries could be reduced in this group of patients who are pCRs or near pCRs and whose PAM/US imaging scores provided by reader team classify them as responders. PAM/US-model outputs will be computed for each patient and compared with reader's assessment . The investigators will also correlate MRI/DWI findings with PAM/US and document agreement/discordance with explanations.

As an exploratory part of this trial, a subset of eligible patients of group #2 of will be selected to undergo three serial imaging studies: at pre-treatment, halfway through treatment (after completion of one week of radiation followed by 2 months of chemotherapy), and at treatment conclusion (after one week of radiation followed by 4 months of chemotherapy). At the treatment conclusion, those deemed to have residual tumor by standard clinical methods (~65%) will proceed to surgery. For the remaining 35% of patients, with no evidence of active disease, surveillance will be initiated. For these patients, serial imaging examinations will be performed every six months (coordinated with standard examinations). Patients deemed to have a recurrence will undergo surgery.

Once every enrolled patient in this subgroup has completed either surgery or two years of surveillance, the participants will be divided into two groups: Group A (those who underwent surgery) and Group B (those treated solely with nonoperative surveillance). As the primary outcome, Group A will demonstrate the accuracy of photoacoustic imaging. As a secondary outcome, Group A will reveal how the tumor vasculature and %sO2 changes over time, which may provide new insight into how complete response patients respond differently from non-pCR patients and how PAM/US can capture these earlier changes. The investigators will follow up on their SOC results in the clinical system and correlate their PAM/US results during the 2-year surveillance with their five-year disease-free survival.

To identify each patient as a complete or non-complete responder, the investigators will analyze photoacoustic images of all group #2 56 patients serially with the reader team assisted by PAM/US models developed earlier in the trial. These quantitative measures, as well as any qualitative local vasculature and %sO2 changes and US morphological changes, will be recorded at each time point. Standard MRI and endoscopic images will be independently collected and analyzed. Since the novel PAM/US model is in the development stage, the investigators may need to continuously re-train the model with more data collected and test the model from new cases to improve the model performance.

As a final part of this trial, the investigators will explore patterns of post-treatment vascular and %sO2 changes during preoperative treatment and serial surveillance. Tumor angiogenesis in colorectal cancer has been well studied using microvessel density (MVD) as a surrogate marker. While high levels of MVD and vascular endothelial growth factor (VEGF) significantly predict poor survival, little is known about the dynamics of vascular and %sO2 changes and the mechanisms of neovascularization after chemoradiation.

The investigators have observed tumor beds revert from oncologically-disrupted microvascular patterns to highly regular architecture typical of the normal rectum where pCR has occurred, though this mechanism is not well-understood. In contrast, heterogeneous and often microvascular-deficient regions have been found consistently in tumor beds with residual cancer at treatment completion. This novel imaging device could serve as a new platform for investigating the dynamics of vascular and %sO2 changes during and after therapy and for assessing potential rectal cancer recurrence early on. In addition to PAM/US imaging studies, the surgeon will perform biopsies at the central region and edges of the tumor bed before treatment, at treatment completion (if no surgery), and at the point of tumor reoccurrence. For patients with surgery, surgical samples will be used. Specifically, MVD and VEGF expression will be assessed by immunohistochemical staining, and both will be quantitatively correlated with PAM/US findings at the center of the tumor bed and at the tumor edges. The results will help the investigators understand how vasculature changes during treatment and at recurrence are correlated with imaging findings.