In the earliest, most favourable cases, chiefly the malignant polyps (Haggitt 1–3, T1 sm 1 (−2?) NO), a local procedure, e.g. using the transanal endoscopic microsurgery (TEM) technique, is appropriate [34, 35]. The resection should be radical (R0) and no signs of vessel invasion or poor differentiation should be present. If this is not the case or if the tumour infiltrates deeper into the submucosa (Haggit 4, T1 sm (2?–)3) or a T2 tumour, the risk of recurrence because of remaining tumour cells or because of lymph node metastases is too high (≥10 %), and the patient should have postoperative CRT or, more safely, be recommended major (TME) surgery [36]. If the cancer diagnosis is verified in a biopsy, presurgical CRT is preferred if the intent is to perform a local procedure [34, 36]. As an alternative to local surgery, alone or with (preoperative) CRT, local RT (brachytherapy or contact therapy (Papillon technique)) can be used in the most favourable cases. The experiences of these treatments are limited outside specialised centres [37], and more prospective studies are required in order for these techniques to be part of clinical routines.

In the early, favourable cases (cT1–T2, some early cT3, N0 [cT3a(−b) and clear mrf (mrf-) according to MRI], ‘good’ group) above the levators, surgery alone, meaning a sharp radical dissection using the total mesorectal excision (TME) technique, is appropriate, since the risk of local failure is very low [5]. A high cT3ab tumour with limited lymph node metastases (N1) according to MRI may also belong to the good group if mrf-. The role of TME in tumours situated in the upper third of the rectum has been much discussed. To avoid spillage of distal tumour cells, a margin of at least 5 cm distally to the tumour on an unfixed specimen is recommended. Although there are indications from the large randomised trials where short-course RT has been given that this treatment even further reduces local recurrence rates [11, 38, 39], surgery alone is recommended since the addition of preoperative RT results in overtreatment of too many individuals [5, 12]. The balance between the reduction in local recurrence rates and long-term morbidity is intricate.

In the intermediate or ‘bad’ group (most cT3 (cT3(b)c+ without threatened or involved mrf (mrf-) according to MRI), some cT4 (e.g. vaginal or peritoneal involvement only), N+), preoperative RT is recommended followed by TME, since this reduces local recurrence rates. Even in the absence of signs of extramural growth on ultrasound or MRI (cT2) in very low tumours (0–5 cm), preoperative RT may be indicated because the distance to the mesorectal fascia is very small. Twenty-five Gy during one week followed by immediate surgery (<10 days from the first radiation fraction) is a convenient, simple and low-toxic treatment [11, 38–40]. It has been used predominantly in Sweden, the Netherlands and the UK, where several clinical trials revealing its efficacy have been performed. The trials have shown that the risk of local failure in the randomised population selected for later resection, i.e. the intention to treat population, has been reduced by 50–70 % versus surgery alone. More demanding and not proven more effective alternatives are 46–50.4 Gy, 1.8–2 Gy/fraction without or preferably with 5-FU (bolus, continuous infusion or peroral) [18, 19, 21, 22]. The chemotherapy was added to the preoperative radiation primarily based upon extrapolations from the postoperative RT trials in rectal cancer (see below) and other GI cancer trials. As indicated above in the paragraph about different strategies in different parts of the world, two large European trials (FFCD 9203 and EORTC 22921 [21, 22]) however recently showed that the addition of 5-FU improves local control with reduced local failure rates after 5 years. These were 16–17 % in the preoperative RT arms alone and 8–10 % in the CRT arms. In the EORTC trial, the same reduction was seen whether the chemotherapy was given concomitantly or only postoperatively. Two trials (Polish, TROG 01.04) have randomised between preoperative 5 × 5 Gy and preoperative CRT (5-FU + 50.4 Gy) without detecting any statistically significant difference in local recurrence rates, DFS and OS. In the Polish trial [18], local recurrence rates were 11 % in the 5 × 5 arm at 4 years and 16 % in the CRT arm (p = 0.2). The corresponding figures were 7.5 and 4.4 % (p = 0.2) in the Australasian trial [19]. In a third trial (MRC-CR07) [39], preoperative 5 × 5 Gy was randomly compared with postoperative CRT if crm was positive. Local recurrence rates favoured the preoperative arm (5 % vs. 17 %, p < 0.001) [39]. In the MRC-CR07 trial including 1350 patients, DFS was superior in the preoperative arm (hazard ratio, HR, 0.76, p = 0.01), whereas OS did not significantly differ (HR 0.91, p = 0.04).

Whenever possible, preoperative treatment is preferred since it is more effective and less toxic than postoperative treatment [5, 15]. Postoperative chemotherapy has otherwise been extensively used in many countries, including Germany and the USA, since decades. The NIH consensus conference with the follow-up NCI statement in 1990 and 1991 stated that postoperative CRT should be standard treatment in stages II and III. The scientific support for these statements was considered strong and based upon several randomised clinical trials [8, 9]. Some of the trials were small (about 50 patients per treatment arm), the results were not entirely consistent, and it was difficult to firmly establish the role for local control and survival from the different treatment components (the RT, the concomitant chemotherapy with the radiation (CRT) or the chemotherapy given before or after the radiation). Taken together, these treatments during 6 months appear to reduce local recurrences in stages II and III with 50–60 % versus surgery alone [41].

In the most locally advanced, frequently non-resectable cases (cT3 crm+, cT4 with overgrowth to other organs [cT4a according to TNM5, cT4b according to TNM6 and 7, see Table 14.1]), preoperative CRT, 50.4 Gy, 1.8 Gy/fraction with concomitant 5-FU-based therapy should be used [5, 6, 23], followed by radical surgery 6–8 weeks later. In a Nordic randomised trial in locally advanced rectal cancers (cT4NXM0), local control was significantly better after 5 years in the CRT arm (5-FU + 50 Gy) than in the RT only arm (82 % vs. 67 %, p = 0.03). Also, DFS and cancer-specific survival were significantly better in the combined modality arm, whereas OS did not significantly differ (66 % vs. 53 %, p = 0.09) [23]. In very old patients (≥80–85 years) and in patients not fit for CRT, 5 × 5 Gy with a delay of approximately 8 weeks before surgery can be an alternative option [42–45].

Standard preoperative CRT means a dose of 46–50.4 Gy together with 5-FU given either as bolus injections with leucovorin at 6–10 times during the radiation (as in the trials proving that CRT provides better local control than the same RT alone) [6, 21–23], prolonged continuous infusion (likely better than bolus) or oral capecitabine or UFT. Extrapolations from other clinical situations and convenience tell that oral 5-FU is a valid treatment [46, 47]. Combinations of 5-FU or other antifolates with other cytostatics like oxaliplatin or irinotecan or targeted biologic drugs have been extensively explored in phase I–II trials, with claimed more favourable results (more downsizing, higher pathological complete response (pCR) rates) but also more acute toxicity. Several comparative randomised trials using oxaliplatin are ongoing. The initial results of these are not favourable [48–50], and these combinations are still experimental. Neither are the initial results of adding targeted drugs like cetuximab, panitumumab or bevacizumab favourable [51–55], although the first publications are, as usual, optimistic. When cetuximab was added to neoadjuvant oxaliplatin-capecitabine and preoperative CRT in the randomised phase II EXPERT-C trial, more radiological responses were seen in the cetuximab arm (89 % vs. 72 %, p = 0.003) in the KRAS wild-type population (n = 90) [56]. Overall survival was also improved (96 % vs. 81 % at 3 years, p = 0.04).

Since the response to preoperative therapy (5 × 5 Gy with a delay or prolonged CRT to 46–50.4 Gy) may influence prognosis [61–63] and thus subsequent therapy, both the extent of surgery and postoperative chemotherapy, attempts to clinically and pathologically restage the tumours have been made. There is an increasing experience in evaluating tumour response by repeat MRI or PET-CT. Using MRI, decrease in size can be seen as well as increase in fibrosis and mucous degeneration indicating response [64]. Using FDG-PET, decrease in uptake can be seen [65–68]. At present, the knowledge about the relevance of these changes is too uncertain to modify the extent of surgery.

Several systems for pathological tumour regression grading have been used (e.g. by [69–72]). The best (reproducibility, prognostic information, etc.) is not known. The tumours should at least be graded into three groups, complete response (pCR), some (potentially in the future good, moderate and poor) response and no response. The proportion of pCRs, meaning the absence of tumour cells after a given treatment for a certain substage, is influenced by intensity of dissection. A standardisation of the dissection is required if pCR rates should be used as a valid endpoint [73].