Patterns of local recurrence in rectal cancer after a multidisciplinary approach

Abstract

Improvements in surgery and the application of combined approaches to fight rectal cancer have succeeded in reducing the local recurrence (LR) rate and when there is LR it tends to appear later and less often in isolation. Moreover, a subtle change in the distribution of LRs with respect to the pelvis has been observed. In general terms, prior to total mesorectal excision the most common LRs were central types (perianastomotic and anterior) while lateral and posterior forms (presacral) have become more common since the growth in the use of combined treatments. No differences have been reported in the current pattern of LRs as a function of the type of approach used, that is, neo-adjuvant therapies (short-term or long-course radiotherapy, or chemoradiotherapy versus extended lymphadenectomy, though there is a trend towards posterior or presacral LR in patients in the Western world and lateral LR in Asia. Nevertheless, both may arise from the same mechanism. Moreover, as well as the mode of treatment, the type of LR is related to the height of the initial tumor. Nowadays most LRs are related to the advanced nature of the disease. Involvement of the circumferential radial margin and spillage of residual tumor cells from lymphatic leakage in the pelvic side wall are two plausible mechanisms for the genesis of LR. The patterns of pelvic recurrence itself (pelvic subsites) also have important implications for prognosis and are related to the potential success of salvage curative approach. The re-operability for cure and prognosis are generally better for anastomotic and anterior types than for presacral and lateral recurrences. Overall survival after LR diagnosis is lower with radio or chemoradiotherapy plus optimal surgery approaches, compared to optimal surgery alone.

Key Words: Rectal cancer; Local neoplasm recurrence pelvis; Pattern of recurrence multidisciplinary approach

INTRODUCTION

Over the past two decades, there have been improvements in the management of rectal cancer in terms of postoperative death (falling from 10% to 2%), locoregional failure (dropping from 30%-40% to less than 15%), conservative surgery rates (increasing from 20% to 60%) and survival, with advances made in the understanding of the biology of this type of tumor as well as staging and the use of combined therapies[1]. The anatomical and technical basis of tumor recurrence within the pelvis has been extensively investigated by surgeons and pathologists, and this has led to major improvements in surgical therapy. Nonetheless, although surgery remains the mainstay of treatment aimed at achieving locoregional control, nowadays the therapeutic approach to rectal cancer is eminently multidisciplinary.

The key to successful surgery is complete excision of the tumor proximally, distally and around its circumference with sufficient margin of normal tissue (R0 resection). In total mesorectal excision (TME) surgery the rectum and its perirectal lymphatic and fatty tissue (mesorectum) is completely mobilized by sharp dissection, as an intact package surrounded by an undamaged peri-mesorectal layer of proper rectal (visceral) fascia, avoiding spillage and growth of residual tumor cells into the pelvis and subsequent development of a local recurrence (LR). Educational programs aimed at training surgeons in this pelvic dissection technique have demonstrated reproducible results in achieving a reduction of the LR of rectal cancer rate by 40%[2] and even greater when associated with radiotherapy (RT)[3] or neoadjuvant chemoradiotherapy (CRT)[4].

Nevertheless, LR of rectal cancer remains a significant clinical problem, associated with severe morbidity, low rates of success of salvage procedures, and eventual death in the majority of patients[5].

It is important to review the patterns of treatment failure resulting after rectal cancer management. Improvements in surgical and adjuvant therapies may affect not only the likelihood of tumor recurrence in the pelvis but also the pattern of pelvic recurrence itself (i.e. pelvic subsites). Knowledge of the pattern and natural history of LR, the associated risk factors for their development and the mechanism by which they occur may serve as the foundation for efforts to improve the results of multidisciplinary care (i.e. RT field design, suitability of lymphadenectomies, strategy in the follow-up monitoring, etc.).

The aim of this review is to characterize and analyze the pattern of LR today following different curative approaches for rectal cancer, with special emphasis on the correlation between subsites of pelvic recurrences and treatment modalities.

LIMITATIONS OF THIS REVIEW

Many of the studies that report patterns of pelvic recurrence have multiple limitations. Some are outdated or do not give exact anatomical information of the location of recurrent tumors in the pelvis. In particular, the diagnostic procedures, methods of documentation, acknowledgement or confirmation of diagnosis, presence or absence of histology, the use of interval pain, anatomic definitions of the rectum, first site of recurrence and cumulative recurrence data, as well as the definition of the LR itself and other details all affect the analysis of incidence rates, timing, and patterns[6].

Old literature concerning the pattern of local failure in rectal cancer was based on planned or symptomatic reoperations data or autopsy series. Planned “second look” procedures and symptomatic surgery were performed in the pre-TME surgery era and LR data obtained may be outdated. Furthermore, autopsies reveal only the end pattern of failure.

On the other hand, most recent reports are based on clinical or imaging data which can be also misleading as the methods of diagnosing and confirming LR and length of follow-up are not described consistently. The actual rate of pelvic recurrence may be somewhat higher than estimated by these reports, as some studies report only first sites of failure, and pelvic relapse later in the course of disease is not always assessed in patients under palliative chemotherapy for distant metastasis.

Trials of preoperative RT or CRT in resectable rectal cancer are characterized by multiple methodological problems because treatments are combined (RT and surgery) to address a heterogeneous condition (various populations and stages of rectal carcinoma) and to achieve a variety of goals (downstaging and improving resectability, as well as decreasing local and possibly distant recurrences and improving survival).

DEFINITION AND CLASSIFICATION OF LOCAL RECURRENCE

Although by definition the term LR is only applicable when the initial or primary surgery is expected to be curative (no remaining macroscopic evidence of disease locally, that is R0 and R1 according to the UICC (International Union Against Cancer), it must be seen as the further development of tumor cell remnants: there is a close biological similarity between a primary tumor and an LR, in contrast to the situation with corresponding organ metastases[7].

Recurrent rectal cancer may be isolated (local or metastatic) or combined (local and metastasis). Indeed LR can be defined as any tumor located within the pelvis, either alone or in conjunction with metastases[6]. Several authors have classified locoregional pelvic recurrence in order to facilitate treatment and compare outcomes. Specifically, the distinction between localized and diffuse pelvic recurrence is pivotal in defining subsequent management and prognosis.

The Mayo Clinic[8] described recurrence in terms of degree of fixation both in term of site (anterior, sacral, right or left) and number of points of fixation (F0-F3). Wanebo et al[9] proposed a classification based on the UICC TNM system: TR1 and TR2 corresponding to intraluminal LR, either following local excision or at the anastomosis; TR3 corresponding to LR at or around the level of the anastomosis with limited extramural spread and without pelvic fixation; TR4 corresponding to invasion into either adjacent urogenital organs or presacral tissues with tethering but no fixation; and TR5 corresponding to invasion into the sacrum or pelvic side walls. On the other hand, others divided pelvic local failures into just three basic types of recurrence: localized (central), sacral and pelvic sidewall.

The Memorial Sloan Kettering[10] group describe a nomenclature based on the anatomical region of the pelvis that is involved (Figure 1). Accordingly, LR is defined as either: axial, subdivided into anastomotic, mesorectal (residual mesorectum) or perirectal soft tissue within the center of the pelvis or perineum following an abdominoperineal resection (APR); anterior, involving the genitourinary tract; posterior, involving the sacrum and presacral fascia and sacral root sheaths; or lateral, involving the muscles (piriformis, elevator), soft tissue of the pelvic sidewall, lymph nodes, major iliac vessels, sacral nerve plexus and lateral bony pelvis. Lastly, the Dutch TME trial[11] classifies LR based on the same pelvic subsites, although the perineum and anastomotic recurrences are grouped separately. We have found the latter to be the most useful classification, although it does not distinguish between two different origins of the pelvic sidewall involvement itself: true lateral involvement by growth of tumor deposit in lymph nodes along the iliac vessels and continuous extension of tumors of central origin.

Figure 1 Types of pelvic recurrence. C: Central; L: Lateral; P: Posterior.

DIAGNOSIS OF RECURRENT RECTAL CANCER

Patients with recurrent cancer are a heterogeneous group. To establish LR or pelvic disease after definitive resection of rectal cancer[12], most authors accept at least one of the following major criteria: (1) Histological confirmation; (2) Palpable or evident disease with subsequent clinical progress; (3) Clear evidence of bone destruction; and (4) Positive positron meission tomography examination, and at least one of the minor criteria: (1) Progressive enlargement of soft tissue mass on repeated computed tomography (CT) or magnetic resonance (MRI) examination; (2) invasion of adjacent organs; (3) subsequent rise in tumor markers; and (4) typical appearance in endoscopic ultrasound, CT or MRI imaging.

Note that according to these major criteria no patient should be accepted as having pelvic recurrence by diffuse pelvic pain.

RISK FACTORS FOR LR

Many factors affect the risk of local recurrence.

Pathological factors

The pelvic recurrence rate is tumor stage dependent: the more advanced the stage the more likely it is that rectal cancer will recur[13]. Many authors have confirmed the association between advanced UICC or Dukes’ stage and the likelihood of recurrence. Not surprisingly, the extent of invasion beyond the rectum affects recurrence, with an incidence of less than 1% in patients in whom no local extension is noted, compared to between 5% and 10% in patients with moderate spread and 15%-25% in those with more extensive spread[14,15]. The number of positive lymph nodes[16] as well as a positive circumferential resection margin (CRM) also influence both LR and survival[11]. Even with combined treatments, the incidence of LR in patients with one of these risk factors (that is, TMN stage IV, T4 tumor, N2 disease or positive CRM in T3 disease) reaches 20%, compared to less than 5% in patients who do not have these characteristics[11]. It is also clear that the combination of risk factors is also important: in patients at T1-T2 the incidence of LR is 1% with a negative CRM but this rises to 12% for a positive CRM, while for those at T3-T4 it is 15% for a negative CRM but 25% for a positive CRM[11]. In patients undergoing surgery with or without preoperative RT, the combination of CRM and lymph node status has been shown to be a more effective discriminator of prognosis than TNM staging[1,11].

Height of the tumor is also a critical factor as LR is also more likely with tumors in the lower third of the rectum (10%-15%) than in patients with tumors in either the middle third (5%-10%) or upper third (2%-5%)[3,5,11,15]. The risk of LR is also related to the position of the tumor within the circumference of the rectum. In the series of Chan et al[17] the rate of LR was 15% (95% CI, 11-22) for tumors affecting the anterior side of the rectum but was 5.8% (95% CI, 3-11) for other locations. Anterior tumors tend to be more advanced, at least in male patients, and the anterior aspect of the TME dissection more difficult to perform in the narrow male pelvis, presenting a higher risk of LR and death than tumors in other sites[18].

The shape (exophytic versus non-exophytic) of the tumor, the presence or absence of budding, lymphatic, venous or perineural invasion, the presence of obstruction or perforation of the tumor together with the degree of tumor differentiation, and fixity of the tumor, all influence the risk of local recurrence adversely[13,19].

Therapeutic factors

Inadequate removal of the primary tumor is the most important factor determining whether the tumor recurs[20]. In addition to the involvement of the CRM, the plane of surgery achieved, as a measure of the quality of mesorectal excision, has been shown to be an important prognostic factor for LR[21]. Surgeon variability is also a widely studied phenomenon. In multivariate analysis “the surgeon factor” has emerged as a critical treatment-independent variable, and is not only related to the volume of operations performed[22].

The incidence of lateral lymph node involvement has been extensively investigated by Japanese authors. The term ‘‘extended lymphadenectomy’’ (EL) refers to the removal of the lymph nodes in the extra-mesenteric area. Generally rates of node involvement have been reported to be 5%-10%, but are markedly higher in stage III tumors in the lower third of the rectum, with rates of up to 15%-25%[23]. However, it is doubtful that remaining lateral lymph nodes after an R0 resection is the major or only source of tumor regrowth[24].

There is abundant clinical data supporting the importance of the pathologic response and downstaging to preoperative CRT. All patients who developed cancer recurrence in the German trial had positive lymph node involvement post-treatment[25]. A pathological complete response (pCR) or greater than 95% CR in the post-CRT is a predictive factor of low LR rate and good prognosis, and several studies have shown that this is the most important independent prognostic factor in multivariate analysis for disease-free survival[4,26].

While attention was traditionally focused solely on the optimal distal mucosal margins required to achieve an oncologically safe resection, not only have these margins been reduced significantly, but also greater importance has come to be placed on the lateral margins for achieving an Ro resection. Short (but negative) distal margins have repeatedly been found not to be associated with pelvic recurrence[27,28]. The risk is not so much of intramural spread as of intramesorectal spread, and probably the risk for mesorectal tumor deposit is higher in node-positive than in node-negative patients[27].

Other factors that may increase a patient’s risk of LR are related to increasing body mass index[29]. It has been shown that obese men are more likely than normal weight males to develop an LR, and that adiposity is a strong predictor of requiring an APR[30].

SITES OF LOCAL RECURRENCE AFTER SURGERY ALONE

In the late 1970s, areas of failure found on planned reoperation (“second look” procedures) of patients at risk of recurrence, in spite of an initial curative resection, were investigated. The seminal article by Gunderson and Sosin[31] showed that as distant metastases alone were uncommon (7%), LR as the only type of failure occurred in nearly 50% of cases of recurrence and as some component in 92%. They also depicted the pattern of pelvic subsites LR (Table 1), showed that disease relapse rates were indeed related to the degree of bowel wall penetration and the extent of nodal disease, and paved the way for the exploration of radiation therapy.

Table 1 Patterns and rates of recurrences after different approaches.

Authors - year	Treatment	5-year local recurrence (%)			Metastases as a first site of recurrence (%)	Median time to LR
		Total	Isolated	Combined
Pilipshen et al[32] (1968-1976)	Pre-TME surgery	25.5	13.3	12.3	NA	16 mo
Heald et al[34] (1978-1986)	TME surgery	3.7	2.6	1.1	NA	14 mo
Swedish Trial[39] (1987-1990)	Pre-TME surgery/	23.0	13.0	10.0	11	NA
	sRT+ pre-TME	9.0	5.0	4.0	19
Mohiuddin et al[67] (1976-1989)	LRT+ Pre-TME	13.0	6.0	7.0	17	25 mo
German Trial[25] (1995-2002)	CRT+ TME/	6.0	3.0	3.0	20	NA
	TME + CRT	13.0	7.0	6.0	20
British Trial[44] (1998-2005)	sRT +TME/	4.7	2.0	2.7	19	NA
	TME+ selective CRT	11.5	6.1	5.4	21
Dutch Trial[41] (1996-1999)	TME surgery/	11.3	6.6	4.7	17 19.3	18 mo
	sRT+ TME	5.8	2.3	3.5		30 mo
Guillem et al[4] (1988-2002)	CRT+TME	4.3	2.3	2.0	22	23 mo
Yu et al[54] (1989-2001)	CRT+TME	8.3	6.3	2.0	NA	NA
Kim et al[56] (2001-2005)	CRT+TME	8.0	3.7	4.3	19	24 mo central 18 mo lateral
Kusters et al[58] (1993-2002)	Unilateral EL/	15.4 (only T3,T4 tumors)			NA	NA
	Bilateral EL	8.3 (only T3-T4 tumors)
Moriya et al[57] (1982-1991)	Nerve-sparing EL	6.2 (14 for N+ tumors)			12	17 mo

TME: Total mesorectal excision; sRT: Short-term preoperative radiotherapy; LRT: Preoperative long-term radiotherapy; CRT+TME: Pre-operative chemoradiotherapy; TME+CRT: Post-operative chemoradiotherapy; EL: Extended lymphadenectomy; NA: Not available; LR: Local recurrence.

In the era before the adoption of TME, surgery alone was associated with local failures of up to 30%-50%. The classic article of Pilipshen et al[32] describes the pattern of LR at a prestigious institution (Sloan-Kettering) in the pre-TME era, with LR rates of 31% in cases at T3-T4 and 49% in cases at N2. Hruby et al[33] provide a detailed analysis of the sites of LR after undergoing surgery for rectal cancer in a series of 269 patients mainly during the 1980s. As can be seen in Table 1, at that time most LR was axial or central, in the block of fatty tissue surrounding the rectal wall, within or contiguous to the operative site, and appeared 6-16 mo before the appearance of metastasis.

In the 1980s details emerged of the first series of patients treated with TME surgery. Heald et al[34] published an accumulated local recurrence rate of 3.7% in a personal series of 115 cases of “curative” low anterior resections[5]. Ten years later, the same author reported actuarial 5- and 10-year recurrence rates of 6% (95% CI, 2%-10%) and 8% (95% CI, 2%-14%) respectively with TME surgery alone. Another TME pioneer, Enker, published a rate of 4.1% for Dukes’ B and 8.2% for Dukes’ C patients[35].

In the 1990s large reductions in local and distant recurrences were reported with TME and so, even without a proper randomized control trial, the TME resection became the new gold standard of surgery for rectal cancer.

Although regarded as equivalent in the pre-TME surgical era, the superiority in terms of LR of the TME low anterior resection (LAR) over the “traditionally” executed APR soon became clear. “Standard” or “traditional” APR for low rectal cancer is associated with a higher rate of positive CRM (30%-60%) and operative perforation (20%-33%), leading to higher LR and poorer survival rates than with LAR[36]. Recently a more radical excision of the levators and puborectalis muscles, carried out in the prone jack-knife position, has been proposed (“extralevator APR”)[37].

As the surgical techniques worldwide evolved to TME resections, European researchers focused on the delivery of a short course of high-dose preoperative RT, without chemotherapy, after ascertaining that adjuvant RT, both before and after surgery, substantially reduced the risk of LR when biologically effective doses of 30 Gy or more were used[38].

LOCAL FAILURES AFTER SHORT-TERM PREOPERATIVE RT

The Swedish Rectal Cancer Trial[39] randomized 1168 patients to surgery alone or to surgery following a 1-wk of pelvic RT (25 Gy in 5 daily fractions), and showed that not only was the 5-year LR rate significantly improved with preoperative RT (23% vs 9%, among the curatively treated patients) but also the 5-year survival rate significantly improved (58% vs 48%). However, this trial was conducted in the surgical era prior to the adoption of TME.

Syk et al[40] reviewed the incidence and location of LR in a group of 880 patients from Stockholm after the introduction of TME surgery, and half of the group also received short-term preoperative RT. In this study, 42% of LR originated from tumors in the upper rectum, and a majority of these patients had not received RT. In all these cases, the recurrence was at the anastomosis and virtually all had visible signs of residual mesorectal fat. Eighteen percent of the patients had LR involving the lateral wall of the pelvis, but only 6% of the tumors involved sites consistent with recurrence in iliac lymph nodes. The authors concluded that lateral pelvic lymph node metastases are not a major cause of local recurrence after TME, and that partial mesorectal excision may be associated with an increased risk of local recurrence due to presacral and/or pelvic sidewall involvement in the upper rectum.

As surgery improved, Dutch researchers then asked whether preoperative short-term RT would still be beneficial in the setting of TME resection properly executed. In the Dutch TME trial a significant benefit was seen with preoperative RT in patients with TNM stage II and III disease, with the two-year local relapse rates decreasing from 5.6% to 1% and from 15% to 4.3%, respectively[41]. The update of the trial reported in 2007 noted a drop in local relapse from 22% to 11% for stage III patients but no significant reduction for stage II patients, and no difference in distant metastasis rate or 5-year overall survival[42].

Subgroup analysis showed a significant fall in patients with cancer in the lower rectum, with nodal involvement but uninvolved CRM. Although those CRM positive patients who received preoperative RT had a lower LR rate than the group with TME alone, this difference was not statistically significant (9.3% vs 16.4%, P = 0.08). The authors arguably concluded that short-term preoperative RT “hardly compensate” for involved CRM[43].

The Dutch group has also recently published a complete and updated analysis of the pattern of LR and the most likely mechanism of recurrence in the trial[11]. They showed that preoperative RT reduces LR in all subsites. However the appearance of LR was slower in the group who underwent RT (2.6 years vs 1.5 years), and if distant metastases diagnosed within 1 mo of LR diagnosis were also considered to have occurred simultaneously, the rate of combined recurrences was higher in the RT + TME group (74% vs 40% in the TME group). In the TME group the recurrences were predominantly anastomotic and posterior, while in the RT + TME group most were posterior and lateral. The anastomotic recurrences were significantly more common in the TME only group, suggesting that RT is especially effective in preventing anastomotic recurrence. Lateral recurrences represented more than 25% of the total in the RT + TME group and most appeared together with metastasis. After LAR, the recurrences were mostly perianastomotic while after APR they were mostly presacral. Perineal LR were found after APR in the TME only group, but not in the RT + TME group. TME alone in node positive disease resulted in considerable local recurrence when the distal margin was 2 cm or less, while RT resulted in a small number of LRs, except when distal margins were less than 5 mm. In total 17% of the patients had a positive CRM, and of those 17% developed LRs (12% in those with T1-T2 tumors, and 24% in T3-T4 tumors with positive CRM).

Given the higher LR rates with narrow CRM, the question of whether selective postoperative RT could improve outcomes in this setting was addressed by researchers in the United Kingdom. The MRC CR07 trial[44] randomized 1350 patients to TME preceded by short-term RT (25 Gy in 1 wk) vs TME followed by CRT (45 Gy plus 5-fluorouracil) if the CRM was < 1 mm. The results showed that the 5-year local recurrence rate was significantly better in the preoperative RT group (4.7%) than the postoperative CRT group (11.5%). However, in those patients with a positive CRM, the LR rates were not statistically different (16% preoperative vs 23% postoperative).

PATTERN OF PELVIC FAILURE AFTER LONG COURSE RT PLUS CHEMOTHERAPY

Postoperative adjuvant strategies to improve outcomes following rectal resection have mainly been explored in the United States. Indeed, in 1990, after positive trials conducted by the Gastrointestinal Tumor Study Group[45] and the Mayo Clinic/North Central Cancer Treatment Group[46], the NCI issued a statement declaring combined postoperative therapy the new standard of care in this setting[47].

Researchers, however, were questioning whether preoperative combined therapy would be even more beneficial. In the 1990s, data from the Memorial Sloan-Kettering Cancer Center and the MD Anderson Cancer Center accumulated[48] in support of that benefit. Moreover, results from three randomized trials (the Uppsala trial[49], NSABP R03[50] and above all the German CAO/ARO/AIO trial[25]), demonstrated the clear superiority of preoperative RT regimens over postoperative therapy in terms of local control with better compliance to treatment and lower toxicity.

The next step was to test the hypothesis that chemotherapy plus preoperative RT significantly improved local control, tumor downsizing and downstaging compared with RT alone. Two randomized trials compared preoperative RT vs preoperative CRT, the study by the Fédération Francophone de Cancérologie Digestive (FFCD 9203)[51] and the EORTC 22921 trial[52], and similar results were reported. In the latter, the five-year results showed that chemotherapy increased the rate of pCR (14% vs 5.3%), translated into a 3% benefit in terms of sphincter preservation and significantly reduced LR rate from 17% without chemotherapy down to 8% with CRT. Thus chemotherapy, regardless of whether it is administered before or after surgery, confers a significant benefit with respect to local control. The main criticism that can be made of those trials is that TME resections were not uniformly implemented.

On the other hand, the favorable effect of delaying surgery after CRT on downstaging (and possibly also sphincter preservation) was shown in the Lyon R90-01 trial[53].

There is, however, limited data on patterns of relapse in rectal cancer patients treated with TME surgery and CRT. Such information might help determine whether modifications in RT dose or field design are warranted (i.e. a local recurrence after RT may be inside or outside the RT field; recurrence outside the field requires an increase in the size of the field, and recurrence inside the field implies the need for an increase in the total dose). We have identified reports on only four series of patients that contain detailed analysis of the pattern of pelvic recurrence after CRT and TME surgery.

From the MD Anderson Cancer center, Yu et al[54] presented a thorough study attempting to identify subsites of pelvic LR in an effort to correlate sites of relapse on CT images with RT simulation films in 46 rectal cancer patients. Of all the LR, approximately two-thirds were in-field (within the radiation field) recurrences and only one-third were marginal (inside but within 1 cm of the border of the field) or out-of-field (more than 1 cm from the border) recurrences. Of the in-field recurrences, nearly 80% occurred in the low pelvic and presacral regions. Multivariate analysis showed that the risk of in-field LR was significantly associated with pathological N stage, while it was notably not with positive CRM or downstaging. The authors suggested various strategies to improve locoregional control in low pelvic and presacral regions.

Hötch et al[12] published a large-scale multicenter study based in Germany to evaluate pelvic sites of recurrence with special attention to radiation ports. Nearly 80% of LR occurred within the treated volume, in the central pelvis, and the pelvic sidewall structures were involved in fewer than 5% of tumor relapses. They found no significant differences in the incidence of pelvic sidewall involvement between APR and LAR cases, however there was a significant difference in the spread of recurrent tumors in the inferior part of the pelvis.

However, a quite different picture has been reported from Korea, where Kim et al[56] examined the patterns of locoregional recurrences in 366 patients with locally advanced rectal cancer who underwent preoperative CRT and curative TME surgery, and assessed the effect of clinical parameters on lateral pelvic recurrence. Eight percent of the patients had LR, of which around 20% and 80% occurred in central and lateral pelvic areas respectively. Multivariable analysis showed that lateral pelvic recurrence was significantly associated with ypN classification (lymph node status after preoperative CRT) and lateral lymph node size. The authors suggested that lateral lymph node metastasis is a risk factor for LR and could be a potentially curable regional disease rather than a sign of systemic disease. Accordingly, they suggested that patients with lateral lymph node size of > 10 mm and ypN0 or lateral lymph node size of 5 mm and ypN+ are a potential subgroup of patients who might benefit from lateral lymph node dissection.

In tumors of the middle and lower rectum, lateral lymph nodes remain a potential cause of locoregional recurrence after conventional TME because they are not removed. EL has been championed mainly by Asian surgeons, who are internationally renowned for their skills in radical surgery.

PATTERN OF RECURRENCES AFTER EXTENDED LYMPHADENECTOMY AND TME, WITH OR WITHOUT ADJUVANT TREATMENT

Around 40% of patients treated for rectal cancer present with lymph node metastases, which occur along the mesorectal nodal chain, along the inferior mesenteric artery lymph nodes or in the lateral pelvic lymph nodes (along the obturator, internal iliac or medial aspect of the external iliac artery). Whether pelvic sidewall lymph nodes should be considered metastatic disease as suggested by the TNM classification (M1) or part of the regional lymphatics (N3) as outlined in Japanese guidelines that are amenable to curative resection, is a contentious issue. Japanese surgeons have adopted the technique of EL to supplement TME, with the aim of minimizing LR and improving survival. Western surgeons do not use EL regularly, and this might pose a risk of local recurrence in the pelvic sidewall in patients operated on without preoperative RT.

In a recent detailed topological analysis of the pattern of lymphatic spread in 605 cases of rectal cancer, 285 cases (47%) were identified as having lymph node metastases. Of this total, 71.5% were mesenteric, 21.5% were lateral and mesenteric, and only 4.7% were exclusively lateral (so-called skip metastases), while among the cases of lateral metastases slightly more than a third were bilateral. The authors[23] concluded that lateral lymph node status is reflective of overall mesenteric lymph node status and that evidence of lateral lymph node involvement may be an ominous sign of advanced disease with an inherent dismal prognosis.

EL is associated with high degrees of urinary and sexual dysfunction and while it is possible to undertake lateral node dissection with autonomic nerve-sparing surgery (NSEL), there are problems with this surgical technique in terms of worldwide uniformity. Moriya et al[57] have studied the pattern of recurrence after NSEL surgery in 306 patients, of which 14% were in Dukes’ stage C, and found an overall LR rate of 6.2%. Dukes’ A and B patients with relapse had suture-line recurrence. In contrast, in the Dukes’ C group, 70% of the recurrences were in patients who had had involvement of more than 5 lymph nodes and 40% in those in whom there had been lateral spread, the number of mesenteric lymph node metastases being the factor which had the strongest impact on the LR. The authors judged that the LR rate with NSEL is similar to that obtained with conventional EL.

Kusters and Van de Velde[58] reviewed 351 patients operated for rectal carcinoma at or below the peritoneal reflection at the National Cancer Center in Tokyo. Standard TME surgery was performed for T1 and T2 (n = 145), and NSEL was added to TME for T3 and T4 (unilateral = 73; bilateral, when the tumor was located centrally, n = 133). They noted that overall there was lymph node involvement in 42% of cases, and lateral involvement in 10%, with “skip” metastases in 3% (mesorectal nodes negative and lateral nodes positive). Overall the 5-year LR rate was 6.6%, while for node-positive (N+) patients the difference between the uni- and bilateral NSEL (32% vs 14%, respectively) was significant.

On the other hand, studies of EL are observational and the reported outcomes are far from uniform even within series reported from Japan. Moreira et al[24] indicated that EL has no advantages for patients in Dukes’ stages A and B and that for cases in stage C it does not significantly reduce LR rates compared to TME. Other series, from both Western and Asian countries, suggested that despite undergoing EL few patients survive for 5 years or more if carcinoma has spread to the pelvic lymph nodes.

Yano et al[59] proposed selective use of extra-mesenteric nerve-sparing lymphadenectomy for those cases in which lateral node metastasis is detected in the CT scan. These authors reported a high level of sensitivity and accuracy (88%) of CT scans for detecting lateral node metastasis, in marked contrast to their diagnostic accuracy for mesorectal lymph node involvement. This same group has published a recent review of lateral lymph node spread in a Japanese journal, emphasizing in particular the rates (20%) of lateral node involvement in T3-T4 cases of low rectal cancer with positive mesorectal nodes. Similarly, Min et al[60] reserved EL for cases in which high-resolution MRI detected extra-mesenteric lymph node metastasis. These authors found a positive predictive value of MRI of 86.4% and 40% for the lateral and paraaortic nodes respectively, and that the location of the lymph node metastasis was the only prognostic factor for cancer-specific survival, with disease in the paraaortic area indicating a worse prognosis than lateral or mesenteric involvement. This suggestion accords with the recommendation in the Guidelines 2000 for Colon and Rectal Cancer Surgery that dissection should be attempted to remove clinically suspected lateral lymph node disease, as far as is technically feasible[61].

A different approach would be to consider EL in various combinations with RT. In the only randomized control trial of EL (n = 23) versus non-EL (n = 22) after preoperative RT (50 Gy) in both groups, Nagawa and colleagues[62] reported no difference in disease-free survival or local recurrence. However, given the small sample size and the fact that the study did not include patients with lateral pelvic lymph node involvement, no safe conclusion can be drawn on the role of either RT or EL in this particular group of patients. Wanatabe et al[63] in a retrospective non randomized study of four patient groups comparing EL with non-EL using either 50 Gy pre-operative RT or no RT found a 5-year survival advantage in the RT group. In addition, the authors reported no significant survival difference between the patients who had preoperative RT with conventional TME surgery compared with those who had EL without preoperative RT, and concluded that preoperative RT could be an alternative to EL.

A comparative non-randomized study by Kim et al[64] recently reported a higher local recurrence rate in patients with TME plus EL than in those with TME plus postoperative CRT. Among those with stage III lower rectal cancer, they successfully demonstrated that the EL group showed a 2.2-fold increase in local recurrence rate compared to the CRT group. However, the 5-year LR rate in stage III in the EL group was much higher than the rate previously reported by the same Japanese group (16.7% vs 7.4%), therefore either a patient selection bias or a different definition of LR cannot be ruled out in this study.

A recent meta-analysis[65] comparing EL and non-EL TME surgery showed no overall difference in cancer-specific outcomes (5-year survival, 5-year disease-free survival and local or distant recurrence). However, as the authors state, the question of whether EL provides benefits in terms of survival or just local control in a subset of patients with advanced rectal cancer could not be safely answered by this meta-analysis.

PATTERNS OF RECURRENCE AFTER INTRA-OPERATIVE RADIOTHERAPY PLUS EXTENDED SURGERY

Several authors reported the impact of intra-operative radiotherapy (IORT) with or without preoperative external beam irradiation and surgical resection in patients with locally advanced or recurrent cancer. The overall available data on IORT showed a favorable impact on local control and in overall survival for patients resected for cure (R0, R1); However, there is a need for randomized studies of the effect of IORT. From a Dutch national referral center, the pattern of LR in 247 patients with locally advanced rectal carcinoma after IORT including multimodal treatment (preoperative CRT and extended surgery) has been analyzed in detail[55]. The 5-year LR rate was 13.2% (7.5% after R0 resections). The most prominent sites of LR were the presacral (44%) followed by the anterior (21%) subsites and lateral spread accounted for less than 10% of recurrences. Around 50% of the LRs appeared in the IORT field, particularly high rates of infield recurrences being observed after dorsal IORT (75%). The authors hypothesize that migration of remaining tumor cells to the presacral space would explain the occurrence of this LR.

PATTERNS OF RECURRENCE FOLLOWING COMPLETE CLINICAL RESPONSE AFTER CRT

The definitive role of an initially non-surgical approach to treatment following complete clinical response (cCR) after CRT has not yet been determined and no definitive conclusions can be drawn before long-term results concerning LR and distant failure are available.

However, Habr-Gama et al[66] have reported the pattern of recurrence and survival of 99 patients with distal rectal cancer (0-7 cm) and cCR following adjuvant CRT, sustained for at least 12 mo, managed by initial non-operative treatment. They observed 13 recurrences: five endorectal (limited to the rectal wall), seven systemic, one combined (endorectal and distant), and no pelvic recurrence outside the rectal wall was detected. There were no significant clinical differences either between patients with and without recurrence, or these same patients according to the location of the recurrence. Surprisingly, systemic recurrence occurred sooner than LR. The authors suggest that a change in the approach to follow-up monitoring may be necessary.

CLOSING REMARKS

The reported patterns and rates of local recurrence, after the aforementioned range of treatment approaches to rectal cancer, used in isolation and as combined therapies are summarized in Table 1. Overall, combining therapies reduces LR rates, delays the appearance of LR and means that when there is recurrence it is less often isolated than after surgery alone.

Table 2 lists the relative frequency of LRs in various pelvic subsites. In recent years, a subtle change has been observed in the distribution of LRs in terms of location within the pelvis, implying the involvement of a different mechanism in their development. In general terms, in the pre-TME years most recurrences were central, perianastomotic and anterior and since the adoption of combined therapies lateral and posterior (presacral) forms dominate. However, the LR distribution is not only related to therapy modality but also to the height of the tumor. LRs in the upper rectum are relatively rare, but when they do occur are usually perianastomotic, originating in the residual mesorectal fatty tissue (as they are treated with partial mesorectal excision, transecting the rectum at about 5 cm below the tumor), and are comparatively more common when only surgical treatment is used. This suggests that while preoperative RT helps to prevent LRs at all sites, it is especially effective in preventing anastomotic recurrences[67]. Isolated anastomotic recurrences are also seen in select cases of very low rectal cancer treated using intersphincteric resection (ISR)[68]. Surgical technique and attention to distal margin can also play a role in preventing this type of LR. The LR rate after ISR is higher in poorly selected cases of pT3 with no previous RT, due to accidental tumor spillage into the intersphincteric space or positive CRM[69]. A lower LR rate has been reported with stapled coloanal anastomosis than for ISR even in T1-T2 patients[70]. As expected, after transanal endoscopic microsurgery, intramural recurrence is the most common type of LR[71].

Table 2 Relative incidence of sub-site locations of pelvic recurrences.

Authors (yr)	Treatment	Axial or central (%)				Lateral (%)	Other (%)
		Anastomotic(perianastomotic)	Anterior(genitourinary)	Posterior(presacral)	Perineal
Gunderson et al[31]	Pre-TME surgery	-1	40	19	31	10
Pilipshen et al[32] (1968-1976)	Pre-TME surgery	40	10		402	10
Hruby et al[33] (1979-1996)	Pre-TME surgery	21	10.7	47	11	11
Dutch Trial[41] (1996-1999)	TME surgery	24	18	32	5	18	2.5
Dutch Trial[41] (1996-1999)	sRT+ TME	13	16	41	-	25	2.7
Syk et al[40] (1995-1999)	sRT +TME	373	30	10		18
Yu et al[54] (1989-2001)	CRT+ CRT	-	44	28	-	10	18.0
Hötch et al[12] (1998-2001)	RT+TME		604	29		10
Kim et al[56] (2001-2005)	CRT+TME		205			80
Kusters et al[58] (1993-2002)	Unilateral EL	25	-	16	16	406
Kusters et al[58] (1993-2002)	Bilateral EL	20	10	16	16	40

APR: Abdominoperineal resection; Anast: Anastomotic; perin: Perineal; Ant: Anterior.

¹0% were APR;

²Post + perineum;

³Anast + ant;

⁴Perin + ant + anast;

⁵Axial o central;

⁶20% ipsilateral and 20% contralateral. TME: Total mesorectal excision; sRT: Short-term preoperative radiotherapy; EL: Extended lymphadenectomy; CRT: Chemoradiotherapy.

Most local recurrences of mid-rectal cancers treated with RT or CRT + TME are related to the advanced nature of the disease. Tumor height of 5 cm or more is associated with a higher incidence of presacral and lateral LR[72]. If the CRM is found to be positive after CRT, the hazard ratio for LR after surgery is significantly higher than if the CMR is involved when no preoperative CRT has been administered (6.3 vs 2.0), possibly because of selection of a population of tumor cells that are resistant to therapy[73]. LR may also sometimes occur even in the absence of an involved CRM possibly owing to lymphatic spread from the distal rectum to lymph nodes in the pelvic side wall[57]. Unilateral EL (lateral lymph nodes on one side of the pelvis are left intact) result in more LR than bilateral ELs, and it has been suggested that the mechanism for the formation of posterior-lateral recurrences may be the migration of tumor cells through the lateral lymphatic vessels to the presacral space under gravity[55]. This would explain why presacral local recurrence is more common in advanced disease than in limited disease.

The highest rates of positive CRM are found with lower rectum tumors. Indeed, TME surgery is not a universal solution for all rectal carcinomas: in low rectal cancer TME may be insufficient to obtain the desired circumferential clearance because of this lack of mesorectum at the level of the pelvic floor. On the other hand, APR surgery mainly results in perineal and presacral LR, which may be prevented by a wider resection[37].

The pelvic pattern of recurrence itself (i.e. pelvic subsites) also has important prognostic value and is related to the potential success of repeat curative intent surgery[74,75]. The operability for cure and prognosis for anastomotic and anterior recurrence are generally better than for presacral and lateral recurrences[75]. Moreover, the upper sacral/lateral invasive type of LR is often associated with synchronous metastatic disease[74]. The type of pelvic invasion is also closely associated with survival after re-resection[74-76]. Finally, it is worth noting that overall survival after LR diagnosis is lower with RT and CRT+TME approaches, than after TME surgery alone[74].