Introduction

At present, the indication for neoadjuvant treatment for rectal carcinoma usually is based on the clinical assessment of the local extent of the tumour, in particular by T and N classification (T3,4 and/or N+) [1–5]. Unfortunately, this policy inevitably is connected with undesirable overtherapy [6–8]. An alternative is offered by pretherapeutic assessment of the relation of the carcinoma to the plane of surgical excision (mesorectal fascia) as today possible by the modern high-resolution thin-section magnetic resonance imaging (with phased-array surface coils) with high reliability [8–11]. The experience of two institutions with a magnetic resonance imaging (MRI)-based indication for neoadjuvant treatment and use of the circumferential resection margin (CRM) status as early surrogate endpoint will be reported.

Materials and methods

This study reports the pooled experience of the Department of General and Visceral Surgery, Vivantes Klinikum im Friedrichshain, Berlin, Germany, and the Department of General and Abdominal Surgery, Johannes-Gutenberg-University, Mainz, Germany, with rectal carcinoma patients, treated according to a uniform concept of MRI-based indication for neoadjuvant therapy and optimised total mesorectal excision (TME)/partial mesorectal excision (PME) surgery in Berlin between 11/2001 and 10/2005 and in Mainz between 12/2003 and 11/2007. Patients in Berlin have taken part in the MERCURY Study (Magnetic Resonance Imaging and Rectal Cancer European Equivalence Study) evaluating the possibilities of modern magnetic resonance imaging for predicting involvement of CRM [11]. The data of Berlin have in part already been reported by Strassburg et al. [12].

Inclusion criteria were:

  1. 1)

    Carcinoma of the rectum (aboral margin of the tumour within 16 cm from the anal verge when measured from below with a rigid sigmoidoscope), at least invading into the submucosa

  2. 2)

    No emergency presentation (no need for urgent surgery within 48 h of admission)

  3. 3)

    No distant metastasis

  4. 4)

    Pretherapeutic high-resolution thin-section MRI employing pelvic phased-array coils performed

  5. 5)

    Indication for neoadjuvant therapy: (a) fixed or tethered tumour (cT4); (b) mobile tumour in case of CRM positivity on MRI (for definition, see below); (c) mobile low rectal carcinoma (<6 cm from anal verge) with invasion beyond the muscularis propria as assessed by ultrasonography and/or MRI (cT3)

  6. 6)

    Radical surgery by anterior resection with total mesorectal excision for tumours of the middle and lower rectum or partial mesorectal excision for tumours of the upper rectum or abdominoperineal excision (APE) with TME

  7. 7)

    Curative intention of surgery, i.e. complete tumour resection with no remaining residual tumour according to the surgeon’s assessment

  8. 8)

    Histopathological examination with special attention paid to the relation between tumour and circumferential resection margin [13, 14]

In some patients, the indication for neoadjuvant therapy was given; however, it was not applied, e.g. in case of refusal by patient or because of extensive perirectal abscess formation. Such patients have been excluded (three patients in Berlin, seven in Mainz).

For the classification of tumour site, the rectum was subdivided into three parts according to the distance of the lower margin of the tumour from the anal verge (assessed by rigid sigmoidoscopy) [15, 16]: upper rectum, 12 to 16 cm; middle rectum, 6 to <12 cm; lower rectum, <6 cm.

The MRI technique of the MERCURY Study was used [17]. CRM-positive was defined as minimal distance between tumour and mesorectal fascia of 1 mm or less.

The surgical treatment for carcinomas of the upper rectum was high anterior resection with PME. Carcinomas of the middle and lower rectum were treated by anterior resection or APE, both with TME.

As neoadjuvant therapy, concomitant radiochemotherapy (50.4 Gy in 28 fractions through 5 weeks, 5-fluoracil either in weeks 1 and 4 or as continuous infusions on days 1–38) was applied. Surgery followed 4–6 weeks after the end of radiotherapy.

The histological classification and grading followed the rules of WHO [18]. For the assessment of anatomical extent of tumour and stage, the sixth edition of tumour–node–metastases (TNM) [16, 19] was applied.

The histological examination of the circumferential resection margin distinguishes between two categories [13, 20, 21]:

  • pathological CRM (pCRM)-positive, direct carcinomatous involvement of the CRM or minimal distance between tumour and CRM of 1 mm or less

  • pathological CRM-negative, minimal distance between tumour and CRM more than 1 mm

For the evaluation of the CRM status, continuous extensions of the primary tumour, discontinuous extensions (tumour deposits, satellites), lymphatic and venous invasion (tumour cells in the lumen with adherence to the vessel wall and/or vessel wall invasion), perineural invasion and lymph node metastasis were taken into account.

Tumour perforation (spontaneous, iatrogeneous) and incision into/through tumour tissue were registered by the pathologist. The assessment of the quality of mesorectal excision by the pathologist was carried out in a part of specimens only.

Differences in frequencies between categorial variables were tested for statistical significance with the chi-square test or Fisher’s exact test when appropriate. For differences of quantitative variables, the U test (Mann–Whitney) was used. The level for statistical significance was set at p < 0.05. Differences with p = 0.05–0.10 were marked as “trend”.

Results

One hundred eighty-one consecutive patients met the inclusion criteria, 93 (51.4%) from Berlin and 88 (48.6%) from Mainz. Table 1 shows the patient, tumour and treatment characteristics. Significant differences between the two institutions are seen relating to pretherapeutic grading of the primary tumour, pathological stage following neoadjuvant therapy and frequency of partial mesorectal excision. Neoadjuvant treatment was given in both institutions in the same frequency (34%).

Table 1 Patient, tumour and treatment characteristics
Full size table

For all patients, the rate of pCRM-negative resection specimens was 170 of 181 (93.9%). This rate was significantly different between Berlin (92/93 = 99%) and Mainz (78/88 = 89%; p = 0.004).

The rate of pathologically CRM-negative resection specimens for patients following neoadjuvant therapy and for those with primary surgery were 56 of 62 (90%) and 114 of 119 (95.8%), respectively. This difference is statistically not significant (p = 0.190).

Table 2 shows the results of univariate analysis of factors influencing the frequency of pCRM-positive resection specimens. For patients treated by primary surgery, significant factors were sex, surgical procedure and institution; for pT, a trend to a significant difference was seen. For patients treated by neoadjuvant radiochemotherapy, tumour mobility, pathological stage and surgical procedure influenced the frequency of pCRM-positive resection specimens significantly; pretherapeutic clinical stage, ypT and institution showed a trend to significant differences. Because of the small number of terminal events (five and six pCRM-positive resection specimens, respectively) a multivariate analysis could not be carried out.

Table 2 Frequency of pathologically circumferential resection margin-positive resection specimens
Full size table

Discussion

The present treatment of rectal carcinoma is influenced predominantly by the advances in surgical methods (optimised TME surgery) but also by the introduction of multimodal procedures, in particular by neoadjuvant radio-/radiochemotherapy and adjuvant chemotherapy for patients with regional lymph node metastasis.

The indication for the neoadjuvant treatment conventionally is based on the clinical classification of tumour extent according to T and N categories of the Union Internationale Contre le Cancer TNM system as assessed by digital rectal examination, rigid rectosigmoidoscopy, endorectal ultrasonography and sometimes pelvic computed tomography. In this way, neoadjuvant treatment is frequently recommended in case of T3,4 and/or N+ tumours, i.e. for about 70–75% of patients [1–5]. By applying such a policy, certainly a quite considerable part of patients is overtreated, especially if optimised TME surgery is performed. This is valid for the about 15–30% of patients with overstaging by imaging procedures (see review by Junginger et al. [8]). But neoadjuvant treatment has to be considered as overtreatment at least for T1,2 N+ patients [22, 23] and patients with pT3 carcinomas invading perirectal tissue in only limited extension (not more than 5 mm [24]).

As neoadjuvant radiotherapy has acute toxicity, subacute and late adverse effects [25–29] and increased occurrence of secondary malignancies [30] is observed, any unnecessary neoadjuvant therapy should be avoided. Thus, the demand for a more selective indication for neoadjuvant treatment increases [7, 8, 31–34].

One possibility to accomplish this goal has been opened by the modern high-resolution thin-section MRI with phased-array surface coils and imaging the tumour in its true transverse plane [11, 17, 35–38]. This technique allows the assessment of the relation between tumour and mesorectal fascia with high reliability (see overviews by Lahaye et al. [9] and Hermanek et al. [10]). In this way, the possibility of complete tumour resection can be predicted. In cases of potentially involved CRM and, thus, questionable complete resection, in particular if the distance between tumour and mesorectal fascia is 1 mm or less in MRI, neoadjuvant treatment with the target of tumour shrinkage and regression (downsizing, downstaging) seems indicated [8, 31, 39–42].

Up to now, there are no studies comparing T/N-based and MRI-based indication for neoadjuvant treatment. A possibility of evaluation of these two policies is the assessment of pCRM status on the resection specimens. In this respect, today [21] it should be differentiated between:

  • pCRM-negative—minimal distance between tumour and CRM more than 1 mm,

  • pCRM-positive—tumour minimal 1 mm or less from the CRM or directly involving the CRM.

Nagtegaal et al. [43] proposed a 2-mm cutoff between pCRM-positive and pCRM-negative specimens. However, this proposal was based on data obtained after a median follow-up of 35 months and 2-year local recurrence rates only. It could not be confirmed for patients from Leeds, with a substantially longer follow-up and 5-year local recurrence rates [44]. According to a review of the literature by Glynne-Jones et al. [21], the large majority of studies dealing with the CRM status used the ≤1 mm definition of the CRM positivity (91.1%; 7,373 of 8,094 patients). This definition is accepted in a new publication from the Netherlands, too (with Nagtegaal as co-author) [45].

Quirke et al. [46] demonstrated for the first time the correlation between the pCRM status and the crude local recurrence rate after a median follow-up of 23 months and conventional surgery: pCRM-negative, 1 of 38 (3%) vs. pCRM-positive, 11 of 13 (85%; p < 0.001). In the meantime, there are data of more than 5,000 patients published showing the correlation between pCRM status and prognosis (local recurrence, distant metastasis and survival; Table 3). In multivariate analyses, the pCRM status was the strongest predictive factor for local recurrence [44, 50].

Table 3 Prognostic relevance of the pathological CRM status
Full size table

The pCRM status has to be regarded as the most important prognosticator of recurrence and survival. Thus, it is increasingly recommended as early alternative endpoint (intermediate endpoint, surrogate endpoint) [44, 51, 57–59].

From a clinical point of view, the negative predictive value of the pCRM status relating to local recurrence is especially important. It describes the probability that, in case of pCRM-negative resection specimens in the further course, local recurrences will not be observed. According to the data of Table 3, the pooled negative predictive value for crude recurrence is 726/778 = 93.3%, and the estimated negative predictive value for actuarial 5-year local recurrence is 92.5% (n = 4089). This emphasizes that a tumour resection with a negative pCRM has to be the goal of rectal carcinoma treatment.

In the reported pooled patients from Berlin and Mainz, the indication for neoadjuvant treatment has been based on the relation between carcinoma and mesorectal fascia as assessed by MRI. Following this, the frequency of neoadjuvant radiochemot-therapy could be reduced from 79% to 34%. This policy, together with optimised TME surgery, resulted in 93.5% of patients in pCRM-negative resection specimens. With this result a 5-year local recurrence rate of less than 10% and a 5-year overall survival rate of at least 80% can be expected (Table 3).

Our analysis of the pooled data from Berlin and Mainz has a limitation because a multivariate analysis could not be carried out because of the small number of terminal events (five and six pCRM-positive resection specimens, respectively). Thus, the conclusion of our analysis, namely the concept of a selective MRI-based indication of neoadjuvant radiochemotherapy, has to be assessed as hypothesis only that should be confirmed in prospective multi-centre observation studies with quality assurance of surgery, pathology and MRI. Such a study has started recently (contact Prof. Dr. T. Junginger, e-mail junginger@uni-mainz.de).