Watch-and-Wait as a Therapeutic Strategy in Rectal Cancer

There are patients whose comorbidities or performance status preclude any attempt at radical operation, who are instead offered CRT or short-course radiotherapy as an alternative for definitive treatment. Some of them appear to achieve a cCR and are being subsequently monitored, even though salvage radical surgery will never be an option in case of tumour regrowth. These patients form an entirely different entity and should be excluded from trials involving W&W. The essence of active surveillance is in patients who are able to avoid surgery, but to whom the procedure will be considered at a time when we feel we need to intervene.

A large number of patients end up being monitored because they decline radical surgery. Most of the time, they receive long-course CRT with or without local excision, and those who achieve a cCR or nearCR enter a surveillance program. These patients represent a significant proportion of patients in the retrospective studies on W&W. Unfortunately, their baseline characteristics are often poorly recorded, making it difficult to know if they would have normally been offered nCRT on a basis of high-risk tumour at baseline. At present, it is not possible to advise a patient who is facing an APE what the likelihood is that preoperative CRT will succeed in achieving a CR and thus the likely success of a W&W approach. Since the pCR rates have been reported to be approximately 25% at best, it is therefore logical to conclude that additional RT where it is not indicated in the hope of achieving CR may not succeed in up to 75% of patients.

Patients with early-stage low rectal tumours, without any adverse features, amenable to local excision or transanal endoscopic microsurgery should undergo the procedure without neoadjuvant treatment. If histopathology confirms the full excision and initial low risk preoperative staging, there is no indication for further treatment, and this approach provides excellent oncological and functional outcomes. However, when high risk features are present, such as third submucosal layer (sm3) invasion, positive margin, grade 3, lymphovascular invasion, tumour budding or mucinous subtype, and in cases of pT2 tumours, local resection alone is not sufficient. Local and/or regional recurrences can be seen in as much as 20% of cases [11]. The standard treatment for completion is radical resection, and since the tumours are low, in the majority of cases it results in a permanent stoma. In the context of organ preservation, only few studies have included patients with early-stage tumours. Habr-Gama et al. did a retrospective study assessing the outcomes of a W&W strategy on patients with cT2N0 rectal adenocarcinoma, less than 7 cm from the anal verge, following two different regimen of nCRT [12•]. Patients in the ‘extended’ CRT group (54Gy of radiation and 6 cycles of 5-fluorouracil (5FU)-based chemotherapy) were more likely to achieve a cCR compared to the standard group (50.4 Gy and 2 cycles of 5FU-based chemotherapy) (85.7 vs 56.6%, p < 0.001). As mentioned above, in the initial study by Habr-Gama [10], the rate of cCR was 26.8% for the entire cohort, which comprised T3 tumours in 69% of cases. Therefore, achieving a rate of cCR of 85%, or even 56% with a standard regimen of nCRT, is impressive and suggests the potential of early tumours for W&W. Similarly, an Italian group of investigators have cumulated several years of experience in organ preservation in early-stage low rectal cancers. They have reported in 2012 the results of a prospective randomised clinical trial comparing endoluminal locoregional resection (ELRR) with laparoscopic total mesorectal excision (TME) for early-stage cancers following nCRT [13]. Eligible patients had clinically staged T2N0M0 disease (staging modalities including endorectal ultrasonography, sigmoidoscopy with biopsies, pelvic MRI and whole-body CT), located within 6 cm of the anal verge, grade 1 or 2 tumours with a diameter of less than 3 cm. Patients with lymphovascular or perineural invasion were excluded. The patients in the study received upfront nCRT (50.4 Gy in 5 weeks with concomitant 5FU), had restaging investigations 6 weeks after treatment and were then randomised between ELRR and TME. With a median follow-up of 9.6 years, there was no statistically significant difference between the two groups with regards to locoregional recurrences, rate of distant metastases, cancer-specific survival, disease-free survival and overall survival. The rate of pCR with nCRT was approximately 27% (28% in ELRR group, 26% in TME group). Other studies investigated local excision following nCRT in early-stage low rectal cancer and reported higher rates of pCR (49–59%) [14•, 15, 16]. All these studies provide interesting data in favour of nCRT followed by surveillance or local resection in patients with early-stage low cancers, suggesting that these patients might actually be ideal candidates for organ preservation. However, they involve a relatively small number of patients. The STAR-TREC trial (NCT02945566) is ongoing and should help clarify the question. Although promising, this subject raises an ethical issue. Unless there is a threatened surgical resection margin, which is considered as a high risk feature, patients with early-stage rectal cancer are not typically considered for neoadjuvant treatment. Offering upfront CRT to these patients, the decision based purely on clinical staging, is risky. At our institution, the initial staging of early low rectal cancer includes a high-resolution pelvic MRI amongst other standard investigations. The results are discussed in our weekly multidisciplinary meeting where images are reviewed and management options discussed. When the CRM is clear and the tumour amenable to local resection, we tend to favour this approach as first-line intervention and rediscuss the case with the final histopathology report. It has the advantage of providing accurate pathological staging on which to base future management decisions. When further treatment is recommended, we favour discussion with the patients regarding completion radical surgery versus CRT in an attempt at organ preservation with close imaging and clinical surveillance, knowing that the latter is not yet standard. Although we have reported previously on the good oncological outcomes of patients who decline surgery and undergo adjuvant CRT [17], we prefer, if available, participation in research trials.

In rectal cancer, an incomplete resection is associated with a higher risk of local recurrence and significantly worse outcomes [18, 19]. The rate of R1 resection was traditionally reported as high as 20–40% for low rectal tumours, partly explained by the complexity of the surgical technique in such a confined anatomical space [20‐22]. The MERCURY II trial, investigating specifically low rectal tumours, validated an MRI staging classification to document the relationship between the low rectal cancer surgical plane (mrLRP) and the tumour [23••]. The overall rate of pCRM involvement in the study was less than 10%, proving the value of mrLRP radiological assessment. Furthermore, 4 factors were found on multivariate regression to be predictive of pCRM involvement: an unsafe mrLRP, invasion from the tumour of the anterior quadrant of the rectum, a tumour height of less than 4 cm from the anal verge and the presence of extramural vascular invasion on MRI (mrEMVI). With these factors, the study group was able to propose a predicted risk of pCRM involvement. Tumours with none of the bad features mentioned above have a risk of involved pCRM of 1%, whereas the risk is as high as 60% with tumours presenting all four factors. Based on this data, nCRT is justified in patients with high-risk low tumours. However, as emphasised in the MERCURY II study, restaging investigations, including MRI and reassessment of the surgical plane is mandatory after nCRT. Patients who achieve a cCR or nearCR could be offered organ preservation. In patients for whom surgery is inevitable, a ymrLRP MRI assessment is crucial, as an unsafe plane carries a 25% risk of an incomplete resection. These patients might therefore be candidates for intensified treatment, such as further consolidation chemotherapy or surgery beyond TME with excision of the compartments involved by the treated tumour.

Patients with locally advanced rectal cancer typically require nCRT, for its proven benefit on locoregional control [2]. Those who reach a cCR following treatment could be candidates for W&W. We have previously discussed the low rectal tumours, but what about the proximal ones, those tumours for which a low anterior resection with sphincter preservation is feasible? There are some pros and cons of a NOM in these cases. Firstly, avoiding surgery prevents, at least temporarily, the patient from facing the possible consequences of a triple-modality treatment on bowel, urinary and sexual function and its impact on quality of life [24]. The low anterior resection syndrome (LARS), which is characterised by symptoms such as faecal incontinence, emptying difficulties, urgencies and fragmented bowel movements, is a recognised entity and is seen in as much as 90% of patients who undergo an anterior resection [25‐27]. Several studies have found that nCRT is a significant factor of poorer functional outcomes after surgery. Secondly, a radical procedure such as TME has its risks of surgical complications, morbidities and mortality [6]. On the other hand, one could argue that salvage surgery might potentially be more challenging than an upfront procedure. This has been described in anal cancer, where surgery is a salvage strategy and carries a higher risk of complications [28, 29]. The late side effects of chemoradiation, including pelvic fibrosis, can complexify the surgical technique. Another argument against W&W in higher tumours is the unavailability of digital rectal examination (DRE). Tumours over 6–7 cm from the anal verge cannot be reached with the examining finger, and DRE has been proven a helpful tool in tumour response assessment and monitoring. At our institution, we favour imaging modalities in our monitoring protocol, but agree that DRE should not be completely discarded. The response assessment will be further discussed below. Finally, and possibly the main argument against W&W in the proximal tumours, is the lack of data on long-term outcomes, as they have often been excluded from the studies.

In summary, when analysing data on W&W, the reader should pay careful attention to the description of the population included (stage of disease, methods of baseline and treatment response assessments, quality of imaging). The limitations of the data we have so far are related to the lack of accurate staging or standardisation of selection policies for CRT. Figure 1 summarises our current institutional policy for rectal cancer management, including organ preservation strategies.

In the landmark study by the German Rectal Cancer Study Group, in which patients were randomly assigned to preoperative CRT versus postoperative CRT, the pCR rate in the preoperative group was 8% [30]. The treatment administered was 50.4 Gy of radiotherapy, combined with continuous infusion of 5FU. Other studies, including a meta-analysis from the Cochrane group, have reported rates of pCR of 11–16% [31‐33]. Organ preservation using this standard combination of RT and 5FU or Capecitabine has shown good outcomes, which will be discussed further below.

There is very little evidence on W&W following short-course radiotherapy. To our knowledge, the only data available comes from a prospective study of organ preservation (W&W or local excision) in elderly patients [34]. In a cohort of 60 patients (age 70 years or older, cT1-4N0-3M0 rectal adenocarcinoma), 30 were not candidates for chemotherapy and were therefore offered short-course radiotherapy (SCRT) with delayed assessment of the tumour response. The investigators presented the results of this unplanned interim subgroup analysis, in which they evaluated the feasibility of SCRT with delayed evaluation of response in a W&W context. The 30 patients in the SCRT group received 5 × 5 Gy over 1 week, and 4 of them had a further 4 Gy boost in 1 fraction, 1 week after completing their radiotherapy. The median interval between the last fraction and the response assessment was 10.3 weeks. Six patients (20%) achieved a cCR, and all entered a W&W surveillance program. With a median follow-up of 21 months, 1 of those 6 patients experience tumour regrowth, and the management and outcome of this specific patient is not described in the article. The Stockholm III trial randomised patients between SCRT followed by surgery within 1 week, SCRT followed by surgery at 4–8 weeks and finally long-course radiotherapy (LCRT) of 50 Gy followed by surgery within 4–8 weeks [35]. Even though organ preservation was not part of this trial, looking at the pCR rate can be informative when discussing treatment strategies in W&W. No information was given about the pCR rate with long-course RT. Surgery at 1 week after SCRT led to a very low rate of pCR of 1.7%, whereas delaying surgery to 4–8 weeks increased it to 11.8%, which is still relatively low [36]. This could indirectly indicate the superiority of long-course chemoradiation in a W&W program.

In France (Lyon and Nice), Jean-Pierre Gérard and his team have one of the largest experiences in contact x-ray brachytherapy (CXB) (Papillon technique). The results of their studies have provided strong evidence supporting the use of CXB in selected cases, with high rates of organ preservation and low rates of local recurrence [37, 38]. Sun Myint and colleagues also shared their experience with CXB [39]. They reported on the rate of local regrowth in a context of organ preservation, in 200 patients treated at their centre between 2003 and 2012. Their cohort comprised 17 patients with very early tumours, less than 3 cm, who were offered CXB upfront after declining surgery or being deemed non-surgical candidates. The remaining patients were all patients with diverse tumour stages, who received CRT (45Gy with concomitant 5FU or Capecitabine) or RT alone (dose not reported). These patients had restaging investigations 6–8 weeks following treatment, and those with a residual disease of less than 3 cm were offered further RT with CXB of 90 Gy in 3 fractions. Again, all these patients either declined surgery or were considered unfit for radical resection, hence the referral for further treatment with CXB. A total of 144 patients achieved cCR post CXB, and only 16 of them (11%) experienced local regrowth. The authors argue in favour of CXB for tumour boost, as it has shown one of the lowest reported rates of local regrowth, but these results need to be interpreted cautiously. All these tumours had already shown good regression, as only those showing residual disease less than 3 cm were eligible for CXB. A selection bias might have been introduced, the cohort including potentially only highly radiosensitive and biologically favourable tumours. Another study by Dhadda et al. has also shown a low recurrence rate of 12%, with a similar cohort of patients, but the same biases and critics apply [40]. The Lyon R96-02 trial aimed to compare CXB with external beam radiotherapy (EBRT), but the study was running at the end of the 1990s [41]. The EBRT techniques were not as sophisticated as the current standard, the EBRT alone arm did not include concomitant chemotherapy and the dose delivered (39 Gy in 13 fractions) is no longer recommended. The OPERA study (NCT02505750) is a phase III international randomised trial currently recruiting. Patients with cT2-T3a/b tumours, less than 5 cm, are randomised between two techniques of boost delivery following nCRT: EBRT of 9 Gy in 5 fractions, versus CXB of 90 Gy in 3 fractions. Hopefully, it will provide answers regarding methods of boost delivery and dose escalation.

Several groups have studied the intensification or addition of chemotherapy in the neoadjuvant part of the treatment. The most frequently studied, and possibly the most promising, is the addition of oxaliplatin. The CAO/ARO/AIO-04 trial investigated the addition of oxaliplatin, both pre- and postoperatively, to the standard regimen of nCRT [42••]. Patients with cT3-4Nany or cTanyNpositive disease were randomised between two arms. The standard arm received nCRT (50.4 Gy with 5FU) followed by surgery and 4 cycles of bolus 5FU. The interventional arm received the same radiotherapy plus infusional 5FU and oxaliplatin, followed by surgery and 8 cycles of a combination of oxaliplatin, leucovorin and infusional 5FU. The treatment including oxaliplatin was well tolerated, without significantly different side effects than the standard arm. Furthermore, the oxaliplatin arm led to a pCR rate of 17% (104/596 patients), compared to 13% (81/615) in the standard arm (p = 0.031). Many other groups have studied the inclusion of oxaliplatin and, in addition to increased toxicities, none has resulted in a statistically significant improvement of the pCR rate [43‐46]. However, as mentioned by the authors in the German study (CAO/ARO/AIO-04), the toxicity of the chemotherapy regimen in these studies had a significant impact on the compliance, which could explain the pCR rate.

More recently, the FOWARC trial randomised patients with stage II-III rectal cancer between three arms: nCRT (46–50.4 Gy with 5FU) followed by surgery and adjuvant 5FU, the same regimen with the addition of oxaliplatin (mFOLFOX-6) and finally 4–6 cycles of mFOLFOX-6 followed by surgery and a further 6–8 cycles of mFOLFOX-6 [47•]. The rate of pCR in the standard arm, mFOLFOX-6 + radiotherapy arm and mFOLFOX-6 alone arm was, respectively, 14, 27.5 and 6.6%. Understandably, the toxicities were higher in the mFOLFOX-6 + radiotherapy arm, but surprisingly the compliance was as good if not better than the standard arm group. A group from China performed a meta-analysis on the addition of oxaliplatin to the standard 5FU-based regimen of nCRT [48]. It included, amongst others, the studies mentioned above, and found that the regimen including oxaliplatin significantly increased the rate of pCR (RR = 1.24, 95% CI 1.02–1.51; p = 0.03). It also led to an improved disease-free survival and lower rate of distant metastases.

In 2006, the Angelita & Joaquim Gama Institute, in São Paulo, Brazil, changed their routine practice and started offering an ‘extended’ nCRT regimen, which consists of 54 Gy in 30 fractions with concomitant chemotherapy followed by a course of consolidation chemotherapy (5FU and leucovorin for a total of 6 cycles: 3 cycles delivered every 21 days and 3 additional cycles delivered during the resting period after radiotherapy completion). They recently retrospectively compared the outcomes of patients with cT2N0 disease in the standard regimen with the extended regimen and found that the extended regimen, which was well tolerated, led to a significant increase in the cCR rate (56.6 vs 85.7%) [12•]. Therefore, patients who received the extended regimen were more likely to be eligible to a W&W approach.

The Memorial Sloan Kettering Cancer Centre (MSKCC) is currently running a phase II study in which patients are randomised between induction chemotherapy followed by nCRT, and nCRT followed by consolidation chemotherapy (NCT02008656) [49]. Patients with a significant clinical response to treatment are being management with a non-operative strategy.

While some people investigate the intensification of treatment, others are trying to customise it, for example by studying the feasibility of omitting routine use of radiotherapy in locally advanced rectal cancer. The MSKCC has done so in 2014 with a pilot study [50]. Thirty-two patients, with cT2-T3Nany rectal adenocarcinoma received preoperative infusional 5FU, leucovorin and oxaliplatin (FOLFOX) plus bevacizumab. Patients who maintained a stable disease or showed progression proceeded with neoadjuvant radiotherapy prior to radical resection (TME), whereas patients who responded to treatment went straight to surgery (TME). Four patients needed radiotherapy: 2 patients preoperatively and 2 patients postoperatively for positive resection margins. The pCR rate of patients who received chemotherapy alone was 25%. The GEMCAD 0801 trial investigated the safety and efficacy of neoadjuvant chemotherapy alone (capecitabine, oxaliplatin (CAPOX) and bevacizumab) in 46 patients with T3 rectal adenocarcinoma with clear predicted resection margins on MRI, also in that case with a selective use of nCRT for patients with progressive disease only [51, 52]. They achieved similar results as the MSKCC pilot study, with a pCR rate of 20%. Although interesting, this strategy of offering nCRT only to selected patients after an induction course of chemotherapy is not yet an approved treatment option outside of a clinical trial. The small number of patients in both studies is a limiting factor to draw any conclusion. The PROSPECT trial (NCT01515787) is ongoing and should help further clarify the role of neoadjuvant systemic chemotherapy alone. If organ preservation is sought, radiotherapy should currently be included in the treatment strategy.

We know that there is a strong correlation between dose of radiotherapy and tumour response. In rectal cancer, the work of Ann Appelt has helped to quantify it [53]. Her team analysed the data on tumour regression from 222 patients with rectal adenocarcinoma, treated with nCRT, and derived a dose-response relationship. Tumour regression was defined using the Mandard system. Total tumour dose was calculated by adding the brachytherapy component of the treatment to the external radiotherapy dose. ‘Complete response’ was defined as TRG1, and ‘major response’ was defined as TRG1-2. They found a strong correlation between total tumour dose and tumour response, with 92.0 Gy being the dose resulting in 50% of response for TRG1, and 72.1 Gy for TRG1-2.

One way of escalating the RT dose is with endorectal brachytherapy. One of its advantages over external beam RT is that it can deliver high dose of RT to a localised area. The exposure to surrounding healthy tissues is reduced, as the dose decreases with distance from the source. Buckley and colleagues did a systematic review on high-dose-rate brachytherapy (HDRB) in rectal cancer [54]. They investigated the role of HDRB alone, HDRB combined with CRT and HDRB in the context of W&W. The weighted mean rate of pCR with preoperative HDRB alone was 23.8% and with preoperative HDRB combined with CRT, 22.2%. The main limitations of this systematic review were the large variations in patient selection, the lack of long-term outcomes and the significant paucity of data on toxicities. Only one study used HDRB in a context of W&W and reported good outcomes [55•]. The treatment administered included CRT (60 Gy external beam RT with combined tegafur-uracil) followed by a 5-Gy HDRB boost. Seventy-eight per cent of patients achieved cCR, and 25.9% of patients experienced local relapse but all were successfully salvaged. Te Vuong and her team at McGill University have accumulated years of experience in HDRB and have reported good oncological outcomes in selected patients [56, 57]. The CORRECT trial (NCT02017704) is an ongoing randomised study evaluating the effectiveness of HDRB compared to standard nCRT in rectal cancer.

Understandably, it makes sense to aim for dose escalation, knowing that it will likely provide higher tumour regression, but in rectal cancer, caution is advised as the majority of patients will still go to surgery. It therefore might be inappropriate to push the dose above a certain level. Patients could face more complex surgeries, with a higher risk of complications and could potentially have to deal with poorer functional outcomes in the long term.

At the time of response assessment, especially in the context of W&W, many clinicians are concerned about residual disease and routinely proceed with biopsy. Even in the absence of any residual mucosal abnormality, some people will proceed with random biopsies of the treated scar area, with the hope of being able to document a complete response. When a visible abnormality is suspicious for residual disease, some people would also proceed with biopsy, in that case to confirm the presence of tumour and support the indication for surgery. The São Paulo group did a retrospective comparative study to determine the value of biopsies post nCRT [80]. They reported a sensitivity of 50% and a poor negative predictive value of 11%, which might be explained by geographical miss. The specificity and the positive predictive value were both 100%, meaning that when cancer cells were found on the biopsy, it resulted in confirmed residual cancer in the resected specimen in all cases. These numbers are derived from only three biopsies truly negative and therefore should be taken cautiously. Nevertheless, some people will say that a positive biopsy is enough evidence on its own to proceed with surgery. We disagree, and question the justification of surgery on the basis of a few cells discovered in the sample, without any way of proving their viability. The aspect of timing is also crucial here. A biopsy taken at 6 weeks might be completely different than one taken at 12 weeks. We know that tumour regression continues beyond 8 weeks, but what if it is still the case after 12 weeks? That is the reason why we strongly favour the concept of regrowth. A biopsy reflects the situation at a precise time point. In our opinion, monitoring for regrowth, using imaging modalities and clinical examinations, for example, is a more reliable measure as it takes into account the notion of time. If there is evidence of regrowth, than the residual cells have proven their viability and their ability to proliferate. That is when an intervention, in most cases salvage surgery, should be considered, but a randomised trial is needed if such an approach is to be adopted to ensure that patients are not disadvantaged during this wait period through the development of metastatic disease.

In order to maximise the chances of organ preservation, some investigators have described a new entity, the ‘near complete response (nearCR)’. It usually refers to tumours that show a very good response to treatment, but do not fulfil all criteria for a cCR. When a patient presents a nearCR, some people would favour TEM when technically feasible and have described good rates of organ preservation with this approach [59, 62]. Proceeding with TEM avoids the geographical miss that is possible with random biopsies, but timing is still an issue. In a study by Martens et al., patients who achieved a nearCR after nCRT were offered the choice between reassessment in a further 3 months, versus TEM [60]. Thirty-nine patients presented with a nearCR. Fifteen of them opted for TEM, amongst which 9 (60%) had confirmed pCR. All the remaining 24 patients who opted for further reassessment eventually achieved a cCR and entered the W&W program. The CARTS study was a prospective multicentre trial in which patients with early-stage cancer, eligible for TEM, underwent the procedure after nCRT [81]. The aim of the study was to accurately determine the number of patients with minimal residual disease (ypT0-1). Out of 51 patients who completed nCRT (50–50.4 Gy in 25–28 fractions with concomitant capecitabine), 47 experienced significant tumour regression (ycT0-2) and underwent TEM. There were 21 patients showing pCR (ypT0N0), 9 patients with ypT1N0 and 17 patients with ≥ypT2 and/or N+ disease. Twenty-eight per cent of patients who had TEM experienced postoperative complication. On top of the inevitable risk of overtreatment, TEM after CRT may result in poorer functional outcomes [14•, 24, 82•]. TEM should only be considered in a very selective group of patients following nCRT. The 21 patients with pCR in the CARTS study were obviously overtreated, and we believe that some of the patients with residual disease would probably have pCR as well, if allowed more time to achieve it. In the study, TEM was undertaken 8–10 weeks following the last fraction of RT.

The definition of cCR varies between published series on W&W, but the most commonly used is the one proposed by Habr-Gama and colleagues [83]. On endoscopic and clinical evaluation, a whitening of the rectal mucosa can be observed, with or without associated telangiectasia (Fig. 2). The rectal wall can also show loss of pliability. Possible signs of incomplete response are, on endoscopy, the observation of any residual ulcer (deep or superficial, with or without a necrotic centre), significant stenosis and on digital examination the palpation of any residual nodule.

Most, if not all, series on W&W have included DRE and endoscopy as part of the reassessment protocol. We agree that those are helpful tools; however, they are not sufficient on their own. More prominent fibrotic changes could be easily interpreted as residual disease at the site of the primary tumour. Therefore, imaging might be a more robust and less subjective method of assessing response.

Positron emission tomography and CT (PET-CT) is an imaging modality actively studied in the context of treatment response assessment. Given the fact that a metabolic complete response does not indicate pCR in all cases, a number of factors are being looked at to increase the value of PET-CT in a W&W context: a reduction in the maximal standardised uptake value (SUV), the total lesion glycolysis (TLG) and its percentage change and the metabolic tumour volume (MV), for example. In a series by Perez et al., the accuracy of PET-CT to predict cCR was 91% and increased to 96% when combined with clinical assessment [84].

Magnetic resonance imaging (MRI) is a powerful tool to assess tumour response. There has been great interest in two different sequences of MRI: the standard MRI (mostly T2-weigthed sequence to assess changes within the treated tumour) and the diffusion-weighted MRI (DW-MRI). In any case, MRI should be of high quality to enable visualisation of fine details of the tumour spread as well as assessment of tumour response. For the last decade an MR-modified Mandard grading system (mrTRG) has been used to identify fibrosis/residual tumour signal and categorise patients in groups depending on the qualitative changes within the treated tumour [85]. mrTRG has also been proven to correlate with ypT and be a predictor of outcomes in rectal cancer patients [86]. According to mrTRG grading system patients are categorised in 5 groups:

This imaging method of assessment response has been tested in a number of international prospective trials and confirmed to be an independent factor of prognosis and also shows to be a reproducible imaging tool [87]. Furthermore, a recent analysis showed that mrTRG is 10 times more likely to identify patients with complete pathological response when compared with clinical assessment, particularly residual mucosal abnormality [88]. A number of investigators have been particularly interested in the diffusion-weighted (DW) imaging proposing that this method could improve specificity of patients’ selection with complete response. Supporters of the DW imaging approach are suggesting that residual high signal intensity at the site of the treated tumour on high b-values images is indicative of areas of residual tumour and these patients should not be offered deferral of surgery approach. However, results of the Maastricht group work showed that clinical assessment with additional DWI missed 15% of patients with complete pathological response [89]. Currently, there is no evidence to indicate that apparent diffusion coefficient (ADC) measurements improve accuracy in detecting patients with a good response to treatment or that it provides any added value when mrTRG is utilised and studies have not yet been able to validate cut-off values or definitions for good versus poor response using DWI or ADC values that can be prospectively tested against survival outcomes [90‐92].

Results of the Deferral of Surgery trial suggest that use of DWI protocols and PET/CT would have excluded over 30 and 60% of patients respectively with no regrowth for at least 1 year. Therefore, mrTRG appears to be the most sensitive method of finding patients with pCR not compromising the regrowth rates when compared with clinical assessment and DWI, PET/CT findings. This data was presented in an oral abstract at the European Society of Gastrointestinal and Abdominal Radiology (ESGAR) annual meeting in Prague in 2016, and the final report is awaited.

The modalities discussed above, namely clinical assessment, MRI and PET-CT, are the most frequent tools used in a W&W management. Endoscopic ultrasound, although proven valuable for the initial staging of rectal cancer, is unfortunately not the most helpful in tumour reassessment post treatment. The images are often distorted and even an experienced eye may struggle in differentiating residual tumour from chemoradiation change [93•]. Similarly, the standard CT scan, due to poor soft tissue contrast, has shown low accuracy of less than 50% to assess and predict pCR [93•].

The published series on W&W have all used different modalities to assess response post nCRT. The MSKCC has created a three-tiered evaluation plan (The Regression Schema) that includes endoscopy, DRE, T2 weighted MRI and DW-MRI. The findings defined for each modality are used to divide patients between complete responders, near complete responders and incomplete responders. This assessment protocol is currently being tested and validated in the MSKCC phase II multi-institutional trial mentioned earlier [49]. At our institution, we favour mrTRG and are validating this biomarker in the TRIGGER feasibility trial [94••]. In the interventional arm of the trial, the patient is classified as either ‘good’ or ‘poor’ responder, based on mrTRG only. We do not take into account DRE and endoscopy findings at this point. Good responders undertake 12 weeks of consolidation chemotherapy and enter a surveillance program (deferral of surgery arm). Poor responders receive 12 weeks of consolidation chemotherapy before being reassessed with a further MRI. Patients who then fulfil the criteria for good responders enter the same surveillance program. Patients still deemed poor responders are referred for radical surgery. Endoscopy and DRE are used in the surveillance protocol, and their findings are taken into account when there is suspicion of regrowth. The trial flowchart is summarized in Fig. 5.

Rectal cancers represent a group of diverse tumours, with inherent characteristics that can potentially influence the outcome of the treatment. Histology is one of them. The vast majority of tumours are adenocarcinoma, but subtypes include mucinous cancers which come with poorer outcomes and relative resistance to CRT [98]. Also, tumours that macroscopically look similar can express a wide range of different mutations. Furthermore, there is intratumoral heterogeneity, which illustrates the danger of supporting a management strategy and predict pCR on a biopsy sample only. The São Paulo group studied this concept, and by looking at somatic mutations in different fragments of a same tumour, found that the majority of the mutations (60%) were present in only one fragment, with only 27% of mutations being expressed in all fragments [99]. Similarly, two rectal cancer patients with the same overall burden of disease can express different carcinoembryonic antigen (CEA) levels. Several studies have found the CEA level, either prior to any treatment or after nCRT before radical surgery, to be a strong factor of tumour regression and pCR [100‐102]. The location of the tumour within the rectum, its size, its circumferential extent, have all been recognised as possible predictive factors of response [103, 104].

There are many characteristics of the host that could potentially influence the outcome of the treatment. Examples include the immune system, the presence of chronic disorders such as inflammatory bowel disease or diabetes and active cigarette smoking. Medication might be an important confounder, for example immunosuppressor agents like cyclosporin, tacrolimus or even simple corticosteroids. There is not much data on medication and response to treatment, but researchers have found that the use of statins increases tumour regression with CRT [105‐107]. There are at least two phase II trials currently running that investigate the value of adding a statin to nCRT in rectal cancer (NCT02161822 and NCT02569645). Similar to the CEA level, the link between the neutrophil-lymphocyte ratio (NLR) or other inflammatory markers and tumour regression is being studied and is promising [108‐110].

The treatment administered, including the modality, the dose, the delays or interruptions, obviously has an impact on the outcomes. Timing, discussed in details earlier, has also a major influence on the response to treatment. Other possible confounders, on which there is paucity of data, include other interventions, like trauma from multiple biopsies, for example. Could it have an impact on the risk of regrowth?