Optimising response assessment to neoadjuvant therapy in rectal cancer to reduce the incidence of ypT0N0 resection.

Background
Organ preservation after total neo-adjuvant therapy (TNT) with attainment of a complete clinical response (cCR) is a common pathway for the management of locally advanced rectal cancer [1, 2]. A three-tiered approach incorporates digital rectal exam (DRE), endoscopy and pelvic MRI to select patients suitable for watch and wait (WW) facilitating organ preservation [2–5]. In the OPRA study patients were categorised into either those with a complete clinical response (40%) or near complete response (32%) and enter WW or an incomplete response (28%) and recommended resection [6, 7]. The exact protocol of assessment and surveillance is institute-dependent with the method, frequency, modality and intervals between investigations varying.
Failure to achieve a cCR or detection of recurrence on a WW program is an indication for Total Mesorectal Excision (TME). In the OPRA study 8–9% of patients who completed TNT and underwent TME resection were found to have no residual cancer (ypT0N0 resection) [7]. In a study by Flom et al., the reported rate of ypT0N0 resection was up to 24% [8]. This highlights the difficulty with the current assessment strategy to accurately correlate a cCR to a pathological complete response (pCR) on final histopathology (ypT0N0). TME is a major undertaking and is associated with a 25–30% rate of morbidity, particularly anastomotic leak [9–11]. In addition, TME surgery has a significant impact on quality of life (QOL) due to stoma formation, bowel, sexual and urinary dysfunction and low anterior resection syndrome [12–14]. Furthermore, perioperative risks are increased and long-term function is poorer with the addition of neo-adjuvant therapy [15–17].
Misidentification of tumour persistence or recurrence results in overtreatment. This study aims to identify patients who had a pCR after neo-adjuvant therapy and TME for rectal cancer and evaluate the clinical decision-making and patient surgical outcomes.

Method

Study population
Patients receiving neo-adjuvant chemoradiation for rectal cancer from January 2016 to December 2024 were identified from a prospectively maintained radiation oncology database. The electronic medical record was accessed to retrieve relevant medical information. All patients were reviewed in the context of the multidisciplinary team (MDT) which included pathology, radiology, medical oncology, radiation oncology and colorectal surgery.

Neo-adjuvant management
Patients were considered for neo-adjuvant therapy after discussion at the MDT if they had mid or low rectal cancers, defined by location below the peritoneal reflection and/or a distance of 12 cm or less from the anal verge. Tumours suitable for neoadjuvant therapy were ≥ T3, or T1–2 cancers with nodal involvement on staging MRI pelvis or fluorodeoxyglucose positron emission tomography (FDG PET) [7, 18]. Other criteria considered were threatened circumferential resection margin (CRM), extramural vascular invasion (EMVI), lateral pelvic side wall involvement or low rectal tumours in which patient and MDT preference was to attempt to achieve organ preservation. Long course chemoradiation (LC-CRT) consisted of 45–54 Gy in 25 fractions with a daily dose of 1.8–2 Gy using a VMAT protocol [19]. The default chemo-sensitisation protocol for LC-CRT was with oral capecitabine but intravenous fluorouracil was also used at clinician discretion. mFOLFOX6 was administered as neo-adjuvant chemotherapy as part of TNT, administered either before (induction) or after (consolidation) LC-CRT.

Assessment of response to neo-adjuvant therapy

Radiology
High-spatial-resolution T2 and diffusion-weighted pelvic MRI was performed at varying intervals following completion of neoadjuvant therapy. In patients being considered for WW a progress MRI was performed 4–6 weeks after completion of neoadjuvant LC-CRT. Further MRI intervals were individualized according to MDT decision or followed the WW protocol for up to 5 years. MRI images were reviewed at the MDT by a subspecialist colorectal radiologist.
Response evaluation was classified using MRI tumour regression grade (mrTRG) [20]. Complete response is defined as mrTRG 1 or 2. Incomplete response or tumour regrowth was defined as mrTRG 4 or 5. mrTRG 3 was classified as a near complete response on the initial MRI but an incomplete response on any subsequent MRI. It is important to note that over the long-term course of this study there were significant changes in MRI image technique and quality, as well as increased experience with interpretation of post-neoadjuvant treatment MRI.

Digital rectal exam and endoscopy
A complete response on DRE was defined as no residual tumour felt or induration [3]. Near complete response included minor mucosal abnormality, irregularity or persistent induration on DRE. Incomplete response corresponded to palpable firm disease. Mucosal assessment was achieved via endoscopic evaluation with flexible sigmoidoscopy. A cCR was defined as visualisation of a flat white scar with or without telangiectasia. Near complete response was defined as persisting erythema or shallow ulceration. Any nodularity or deep ulceration was considered an incomplete response [3]. Biopsy was performed selectively and at the discretion of the treating colorectal surgeon.

Pathological assessment
Assessment of TME specimens involved a comprehensive protocol. The unopened tumour was fixed for at least 24 h, followed by evaluation of mesorectal excision quality. The CRM was inked, and the unopened rectum was cross-sectionally sliced at 5 mm intervals. Specimen slices were then oriented craniocaudally and photographed. The tumour bed was entirely embedded, correlating macroscopic, operative and radiological findings, with sampling of adjacent organs when present. Lymph node dissection aimed for a minimum of 12 nodes. If no tumour was identified in the original sectioning, much of the remaining rectum was embedded such that, at a minimum, changes such as fibrosis related to neoadjuvant therapy were confirmed in the tumour bed.

Watch and wait protocol
Patients entered a WW program at the discretion of the MDT which began on completion of neo-adjuvant therapy and lasted 5 years. The surveillance strategy involves 3 monthly clinical reviews with DRE for the first 2 years, then 6 monthly for 3 years. 3 monthly alternative MRI or endoscopy for the first 2 years is followed by alternative MRI and endoscopy every 6 months for years 3 to 5. Patients are reviewed by the MDT every 6 months. In addition, patients followed the standard surveillance protocol for patients with colorectal cancer involving CEA 3–6 monthly, yearly CT chest, abdomen and pelvis or FDG-PET and colonoscopy at 12 months.

Results

Patient cohort

Neo-adjuvant treatment approaches
During the study period 110 patients received neoadjuvant chemoradiation, with treatment approaches evolving significantly over time. From 2016 to 2020, the institutional preference was for LC-CRT followed by routine TME 8–12 weeks later (Fig. 1a). A small number of patients who achieved a cCR entered a WW program (Fig. 1b). After 2021 patients who would previously have been offered LC-CRT prior to low/ultralow anterior resection (LAR) or abdominoperineal resection (APR) were instead offered TNT. Patients then proceeded either to WW (if they had a cCR) or TME (incomplete CR or patient preference) (Fig. 1a, b). Of the total cohort, 67 patients (61%) were treated with LC-CRT and 43 patients (39%) underwent TNT with 18 (16%) treated with induction and 25 (22%) treated with consolidation chemotherapy. A WW strategy was pursued for 23 patients (21%), with the majority of these cases occurring after 2019 (Fig. 1b). Notably, in 2023 and 2024, 30–40% of TNT cases proceeded to a WW approach, reflecting a significant shift in clinical management (Fig. 1b).

Outcomes of neo-adjuvant treatment
Of the 110 patients, 78 (70%) underwent TME (Fig. 2). This group included 71 (91%) planned resections for a failure to achieve a cCR and 7 (9%) salvage resections. Among the 32 (30%) who did not undergo a TME, 23 (21%) entered the watch and wait pathway. With in this cohort, there were 7 salvage resections and 1 patient death from urosepsis. Fourteen patients remain on or have completed the WW program, corresponding to a 64% WW success rate. In 20 patients, TME was deemed unsuitable for the following reasons: progressed or failed to respond to neo-adjuvant treatment (n = 8), underwent pelvic exenteration (n = 5), declined surgery (n = 3), unfit for an operation (n = 2), local excision (n = 1) and chemotherapy-related mortality (n = 1) (Fig. 2).
Twelve cases (17%) of ypT0N0 status were identified: eight in the LC-CRT group and four in the TNT group (Fig. 2). This included 11 planned resections and one salvage resection.
Two patients in the ypT0N0 group died from causes unrelated to cancer; there have been no recurrences in ypT0N0 patients. Although numbers are small, cancer-specific survival appears equivalent between patients managed on the WW program and those with ypT0N0 resection, No cancer-related deaths have been recorded in either group.

ypT0N0 cases

Patient characteristics
The mean age of the 12 patients was 65 with the majority being male (Table 1). Tumours were mostly mid-rectal, T3, N0 or N1 and EMVI and pelvic side wall node negative (Table 1). Eight patients had neo-adjuvant LC-CRT in the period from 2016–2020 with the exception of one patient in 2023. Four patients were treated with consolidation TNT. Figure 3 demonstrates an example of a ypT0N0 case. This elderly male was diagnosed with a low rectal tumour (Fig. 3a, b, c) and proceeded to have consolidation TNT. After a prolonged assessment period he failed to achieve a cCR on MRI (Fig. 3d, e) but had a cCR on endoscopic assessment (Fig. 3f). He proceeded to an APR which was complicated by delirium, ileus, rapid atrial fibrillation and an aspiration pneumonia requiring an intensive care admission. Pathology showed cPR with amyloidosis which was thought to be a confounding factor for ongoing tumour signal on MRI which prompted surgery (Fig. 3h, i).

Discordant investigations for ypT0N0 Cases
In the LC-CRT group 5/8 ypT0N0 patients were committed to a planned resection based on MDT decision prior to neoadjuvant therapy (Table 2). One patient entered WW but there was concern for recurrence on endoscopy resulting in a salvage TME. Two patients were considered for WW but were determined to have failed to achieve a cCR due to apparent disease persistence on MRI in one, and endoscopy in the other. In the TNT group three patients had an incomplete response according to MRI with persistent tumour signal (Table 2). All three of these patients had no evidence of tumour on endoscopy and DRE. The remaining patient had evidence of tumour on MRI, endoscopy and DRE but had a negative biopsy. Overall tumour persistence was misidentified in 10/11 (91%) on MRI, 3/6 (50%) on endoscopy, 3/9 (33%) on DREs, and 2/3 (66%) of patients with a biopsy. Discordance between investigations was common with all patients in the TNT group having discordant investigations (Table 2). In the LC-CRT 3/6 patients who had multimodality pre-operative assessment had discordant investigations. The sensitivity and specificity of MDT detection of cCR were 61% and 95% respectively. The positive and negative predictive values were 86% and 82% respectively.

Surgical outcomes
Patients were generally high risk with 7/12 being American Society of Anaesthesiologists 3 [21]. The majority were male patients with low rectal tumours (Table 3). Operations included four ULAR, seven APRs and one trans-anal TME. All patients treated with ULAR had covering ileostomies. Length of stay was significant with 10 patients requiring a 10–30 day admission and one patient staying > 50 days (Table 3). Surgical morbidity was high with all patients having a complication. There was no surgical mortality. Six major complications as defined by the Clavien-Dindo system occurred, including two anastomotic leaks [22] (Table 3).

Discussion
In the era of TNT there is a delicate balance to be struck between overtreatment and undertreatment of rectal cancer. The difficulty is in achieving a balance between the attainment of the two main aims in rectal cancer: survival and quality of life.
Overtreatment from misidentification of tumour persistence or recurrence has significant consequences. Surgery risks morbidity from complications including anastomotic leak, wound infections and lengthy hospital admissions. QOL is impacted from stoma formation, deconditioning, bowel, bladder, sexual dysfunction and low anterior resection syndrome. Conversely, undertreatment risks disease progression with local and/or distant failure and missed opportunities for curative intervention. QOL may be impacted by the psychological distress from frequent surveillance. The key challenge is developing an assessment strategy capable of navigating this complex area of decision making with precision.
Assessment occurs in the context of an MDT and shared decision making with the patient. Decisions are based on clinical information gained from local staging with pelvic MRI, endoscopy, DRE, biopsy and systemic staging with CT, FDG-PET and CEA. Each modality has limitations and tends to have high specificity and lower sensitivity which shifts the balance of treatment decision to over-treatment, mitigating the risk of missing patients with persistent cancer. The decision making by MDT is dependent on the accuracy of investigations and consequently is specific (95%) but has low sensitivity (61%).
MRI has a low sensitivity for identifying cCR at 36% but a high specificity of 93% [24]. High-quality MRI sequencing with diffusion restriction and interpretation by expert radiologists is essential to maximizing diagnostic accuracy [20, 23]. The most accurate method for assessment is the use of the mrTRG grade score which incorporates DWI with interpretation by expert radiologists [23]. The split scar sign is an inconsistent finding and is limited in applicability in cases of complete response when no fibrosis is visible. It is also highly dependent on MRI quality [23]. In this cohort 10/11 patients had concerns for residual tumour on MRI assessment which influenced the MDT decision to recommend surgery.
Time lapse from radiation to MRI assessment is important for accuracy with a > 10-week interval from radiation therapy recommended [26]. In this cohort of TNT ypT0N0 cases the interval between radiation therapy and pre-operative MRI was > 12 weeks for the 3 cases of reported incomplete response. FDG-PET can be used to supplement MRI findings; however, it has not been proven to be a superior imaging modality for assessing local response to neo-adjuvant therapy [27, 28]. Overall, MRI remains useful as a powerful investigation for identifying potential candidates for organ preservation who require further clinical assessment [20, 23].
Endoscopic assessment was developed from Habr-Gama’s initial work in which clinical response was classified depending on the visualisation of a flat scar, ulcer, nodule or stenosis [3]. Endoscopic identification of a flat white scar is most predictive of a cCR with a positive predictive value of 70–80% [29]. Confidence of a pCR decreases to 50% if a small ulcer is present [29]. Overall, the sensitivity and specificity for detecting a cCR with endoscopy are 65% and 95% [30]. Use of chromoendoscopy and assessment of pit pattern may improve specificity [31, 32]. AI image assessment has shown promise being as capable of reliable assessment as experienced clinicians but has not shown superiority [33].
Endoscopic biopsy is limited by sampling error and offers minor clinical benefit due to a high false negative rate with a sensitivity to detect malignancy of only 12.9% [34]. Mucosal biopsy is highly prone to sampling error due to lateral scatter of tumour cells with 50% of specimens having viable cells outside the visible ulcer margin [35]. Endoscopic biopsy is limited to sampling the mucosa and submucosa. Tumour cells are most likely found at the deepest edge of the invasive front. In patients with T2–4 tumours only 13% have cells detectable in the mucosa, 56% in the submucosa and 98% in the muscularis propria [36]. Unfortunately US-guided biopsy does not increase the accuracy of biopsy [34].
DRE is routinely performed for rectal tumours, often at the time of endoscopy. Tactile feedback may give further useful information if in reach of an experienced finger [4]. The examination however, underestimates clinical response with a sensitivity of 24% and a specificity of 56% [37].
In the current methodology of the assessment of local tumour response to neoadjuvant therapy MRI, endoscopy and DRE all underestimate the effect of treatment, risking complete responders being misclassified as having residual or recurrent tumour. DRE has a low sensitivity and specificity. Biopsy over-estimates response, as it frequently will miss the presence of deeper malignant cells. In many cases discordant results between investigations occur. In this case series discordance was universal for the TNT cohort and common in the LC-CRT cohort. When faced with discordance there are three options.
The first option is to watch and wait longer with sequential interval assessment. Further regression or growth will occur, given enough time the situation will declare itself [38]. However, there is concern with this strategy that distant metastatic disease may seed from a persistent primary. The second is to confirm recurrence with the acquisition of tissue. This may be via partial (submucosal) or full-thickness (including muscularis propria) endoscopic or transanal excision. The advantage of a submucosal biopsy is that this method does not disturb the TME plane. This approach may be more suitable for T2/T3 tumours with a low risk of missing skip lesions [39]. Full-thickness biopsy will provide a more accurate diagnosis for T3/4 tumours but risks disrupting the TME plane [3]. Recent trials show no difference in oncological outcomes for patients undergoing local excision vs TME for low-risk rectal tumours with improved QOL and reduced morbidity [40–43]. Further information will be available pending the short and long-term results of a number of randomised control trials [44, 45]. The third option is to proceed with a TME accepting the 25–30% risk of morbidity and 1.3% risk of mortality [11].
Accordingly, patients may be categorised based on MRI, endoscopy, DRE and biopsy results into those with a concordant complete response and proceed with WW or a concordant incomplete response and proceed with surgery. Patients with discordant results may be considered for short interval surveillance, TME resection or local excision, with the decision for WW or TME dependent on further pathological results. A prospective randomised clinical trial to assess the optimal management strategy for patients with discordant results would add greatly to clinical decision making.
Future options to improve assessment may include introducing new modalities to the assessment process such as a liquid biopsy or biomarkers. To date circulating tumour DNA (cTDNA) has not been shown to have a significant correlation with pCR but its utility is the subject of further clinical trials [46, 47]. The ideal biomarker would be a highly sensitive and specific liquid biopsy and in the era of personalised medicine, based on the genetic mutations identified from the initial biopsy. Development of biomarkers including molecular, metabolic and haematological is currently the topic of avid research [48–50]. New imaging modalities combining anatomical and functional capabilities such as Ga-FAPI-04 PET may show promise [51]. Alternatively, enhancing the accuracy of assessing current and future radiological, clinicopathological and molecular markers using radiomic AI to enhance predication models may assist clinical decision making [52, 53].
This cohort study has a number of limitations. Long-term oncological and QOL outcomes are a question of interest comparing patients who did and did not undergo resections for ypT0N0 disease. Crean et al. found in their retrospective study in patients treated with LC-CRT a higher mortality but better oncological outcomes in the ypT0N0 group compared to the WW group [16]. Surgical morbidity was also high with 58% experiencing an acute complication [16]. The high surgical morbidity in this group is not unexpected and should be interpreted with consideration of the operative risk factors including all patients having had radiation treatment, a male predominance and low tumours. Morbidity in this cohort is also not balanced by simple resections without pre-treatment.
This study has relatively low numbers and is the experience of a single institution. Furthermore, the study period (2016–2024) coincided with significant changes in pelvic MRI and endoscopy technique, quality and experience. Treatment strategy also transitioned from LC-CRT to TNT and WW. With current experience and techniques, the incidence of ypT0N0 resection may be lower than 12%. In this retrospective study there was a small number of ypT0N0 cases which limits statistical power and generalisability. Future benefit would be gained from a prospective multicentre cohort study to compare survival, QOL and cost in ypT0N0 patients treated with TNT vs patients who remain on a WW program.

Conclusion
Misidentification of tumour persistence or recurrence risks overtreatment of rectal cancer. Multimodal assessment with MRI, flexible sigmoidoscopy, DRE and biopsy is useful because the results of these investigations categorize patients into those with concordant and discordant results. Patients with discordant results are at a greater risk of an ypT0N0 resection, excluding these patients from an opportunity for organ preservation. Alternative strategies, such as short interval surveillance, local excision or resection with careful patient counselling can be considered in this group. Prospective validation of which management option is optimal for patients with discordant investigations after neo-adjuvant therapy for rectal cancer would assist greatly in clinical decision making.