While the above-mentioned studies all suggest that most patients with cCR after neoadjuvant chemotherapy can achieve prolonged local tumor control without surgery, a number of questions must be answered before NOM can be considered a standard option for patients with LARC.

The proportion of patients responding completely to neaodjuvant chemoradiation seems small and the optimal time to assess clinical response unknown. Tumor response depends on radiation dose, but doses beyond 54 Gy are rarely used in LARC patients. Adding other drugs effective in colon cancer as radiosensitizers beyond fluorpyrimidines has been found to be ineffective or prohibitively toxic [5–8]. Tumor response to chemoradiation is closely associated with time, and in patients undergoing TME after nCRT, the proportion of tumors with pCR increases with the time interval between chemoradiation and surgery [9]. As prolonging the interval to surgery and postoperative systemic chemotherapy may be unsafe in patients at risk of LR, attempts have been made to deliver systemic chemotherapy immediately before or after chemoradiation [10]. Delivering systemic chemotherapy before rather than after surgery has been shown to increase tumor response without delaying the treatment of potential micrometastatic disease. In these patients, the assessment of the clinical response, with the potential recommendation of NOM or surgery, is performed at the completion of both chemoradiation and systemic chemotherapy [10, 11]. This approach has resulted in pCR rates as high as 38 % in patients with clinical stage II and III disease and has the added advantage of increasing compliance with adjuvant systemic chemotherapy as well as shortening ileostomy time for patients after low anterior resection [10].

The lack of a reliable and uniform method of distinguishing post-treatment scar from residual tumor in the bowel wall or regional lymph nodes is the main obstacle to NOM in patients treated with neoadjuvant therapy. Most authors agree that digital rectal examination, endoscopy, and imaging studies should be used (Table 19.1). A flat white scar with or without telangiectasia and a normal digital exam are good predictors of pCR, while the presence of superficial ulceration or a palpable modularity on digital rectal exam considered an indicator of incomplete response [12, 13]. While clinical assessment tends to underestimate tumor response, there is always a possibility that tumors are concealed in or behind an apparently normal scar in the rectal wall [14]. Endorectal ultrasound, computed tomography (CT), and positron emission tomography with [¹⁸F]fludeoxyglucose provide a rough estimate of tumor regression but are not sensitive enough to identify pCR [15]. Conventional MRI morphological sequences (e.g. T2- and T1-weighted images) cannot differentiate residual tumor from surrounding fibrosis, but diffusion-weighted (DW) MRI sequences may improve the diagnostic performance of morphological MRI sequences in differentiating pCR from residual tumor [16]. The criteria used to grade response undoubtedly influence the observed clinical outcomes: a strict definition reduces the proportion eligible but increases the chance of NOM success, while looser criteria increase the number of eligible patients but also risk of local tumor regrowth and distant metastasis. Currently, there are no validated criteria defining clinical and radiological tumor response, but a new set of criteria categorizing response in a 3-tier system is currently being tested in a prospective clinical trial [17].

A number of patients with apparent cCR develop tumor regrowth during follow-up. As most regrowth occurs in the bowel wall, repeated endoscopic exams are essential. Any suspicious changes in the scar should be biopsied. MRI should also be performed regularly to detect nodal disease. Changes in the size, contour, heterogeneity, or restriction of diffusion should raise the possibility of relapse. Repeated exams and continuous monitoring are often necessary to confirm recurrence.

Ultimately, finding reliable predictors of response to neoadjuvant therapy would help identify patients most likely to benefit from NOM and reduce toxicity for those who will likely have poor response. Tumor size and stage seem to predict response, with smaller, early-stage tumors being more likely to yield pCR. The search for molecular predictors of tumor response has not yielded any breakthrough findings so far. We have previously shown that rectal tumors with a KRAS mutation are less likely to respond to nCRT [18]. However, these findings await validation by studies of large independent cohorts.

Unfortunately, there is no level 1 evidence regarding the oncological and functional outcomes of NOM versus standard TME after a cCR. Ideally, a randomized study should be performed, with a non-inferiority design for the non-operative arm. However, there are 2 reasons why this kind of study is difficult to perform. First, a non-inferiority study requires investigators to demonstrate that survival will not be compromised in NOM. Such a study requires a large sample size that will be difficult to achieve. Second, it is unlikely that patients who are told that NOM is an alternative option potentially offering similar oncological results would opt for randomization with a chance of undergoing surgery anyway.

Meta-analyses are also not available. Thus, the only types of studies we can analyze are retrospective series or prospectively followed patient series. Our search terms on PubMed were “complete response,” “rectal cancer,” “non-operative management,” “watch and wait,” and “wait and see.” We will discuss the oncological and functional results in the next chapters.

In this overview, we included studies in which patients with a cCR as established by digital rectal examination, endoscopy, and MRI were compared to a cohort of patients who had a resection and demonstrated pCR on pathologic examination. There is also one study in which patients were managed by NOM after cCR diagnosis established by MRI alone. In our opinion this is not the standard of care, so we did not include this study [19]. Table 19.2 shows the oncological outcomes of the 5 comparative studies in order of publication.