Prognostic Determinants of Patients With Lateral Nodal Involvement by Rectal Cancer

Abstract

Objective

To clarify the characteristics related to long-term survival in patients with lateral nodal involvement.

Summary Background Data

Few reports have addressed the prognostic determinants in patients with actual lateral nodal involvement, which are important in determining treatment.

Methods

Review of a prospective colorectal database at a single institution for a 10-year period (1987–1996) identified 53 patients with lateral nodal involvement.

Results

All nine patients who underwent resection of synchronous distant metastases developed recurrence and died within 3 years. Of the 44 patients without distant metastases, 25 (57%) developed locoregional recurrence, and the overall 5-year survival rate was 32%. Multivariate analysis showed that age, total number of involved nodes (mesorectal and lateral), and circumferential surgical margin involvement had independently predicted postoperative survival. Patients with three or fewer nodes involved accounted for one third of lateral-positive patients, with a 5-year survival rate of 75%, whereas the 18 patients with four or more involved nodes had a 5-year survival rate of 4%. All eight patients with circumferential margin involvement died of carcinoma, and seven developed locoregional recurrences. Involvement of other pelvic organs had no effect on prognosis, nor were adverse prognostic outcomes noted by the region of lateral involvement.

Conclusions

For patients with lateral involvement, the most important prognostic variables are distant metastases, the total number of nodes involved, circumferential margin involvement, and age. Selection of patients based on these variables may lead to the identification of a subgroup for whom lateral nodal dissection could be the first treatment choice.

Locoregional recurrence after curative surgery is common in patients with rectal cancer. Total mesorectal excision has been reported to reduce the incidence of local recurrence. ^1,2 Conversely, inadequate mesorectal excision, characterized by the presence of tumor in the radical resection margin, results in an increased incidence of local recurrence and a decreased likelihood of survival. ³ However, some patients have extramesorectal spread to the lateral pelvic nodes that cannot be resected by the total mesorectal excision procedure; this might explain why in some patients local recurrence develops even in the absence of circumferential margin involvement. ⁴

In the TNM classification, a lateral lymph node is defined as nonregional for rectal cancer, and it has been widely accepted since the 1950s that the survival rate of patients with lateral involvement is unfavorable. ^5–7 The incidence of lateral cancer spread in patients with lower rectal cancer has been reported as 10% to 25%. ^6,8 Further, occult metastases are present in approximately 4% of patients who, based on conventional histologic examination, are negative for lateral node involvement. ⁹ It may safely be said that there have never been any commonly accepted treatments for patients with such disease. Some favorable prognostic results after systematic lateral dissection have been reported, ^10–12 but other reports have discouraged the use of this procedure from several points of view, including prognostic outcome (lateral spread is often regarded as a form of systemic disease), functional outcome (radical lateral dissection causes sexual and urinary dysfunction), and study design (no randomized controlled trials have been done to verify the efficacy of this procedure). ^13–15

To date, there have been few detailed reports on the prognostic outcome of patients with actual lateral nodal involvement, although such studies should be the basis for deciding treatment strategies in these cases. In this paper, we clarify the factors that determine the prognosis of patients after lateral nodal dissection, based on the data of patients with proven lateral node involvement.

PATIENTS AND METHODS

Patients

During the 10-year period between 1987 and 1996, 545 patients with advanced rectal cancer, including 39 with distant metastases, underwent potentially curative surgery at the National Defense Medical College Hospital in Japan. Among these 545 patients, 250 who had no excessive surgical risk underwent systematic lateral dissection, and 53 (21.2%) had lateral nodal involvement pathologically. Lateral dissections were performed in patient with lower rectal cancer whose tumor penetrated the rectal wall on the basis of preoperative images (barium enema, endoscopy, and endoluminal ultrasonography), digital examination, or intraoperative manipulation, or whose tumor was limited to the muscular layers but whose nodal involvement was evidenced by the preoperative images (computed tomography and/or magnetic resonance imaging and endoluminal ultrasonography) and intraoperative palpation. In addition, lateral dissections were performed for patients who had cancer above the peritoneal reflection and whose lateral nodes were suspected to be involved based on preoperative images (computed tomography and/or magnetic resonance imaging) or intraoperative examination.

These 53 patients comprised 33 men and 20 women with an average age of 56.5 years (range 25–85). The location of the tumor anal edge was the upper rectum (above the peritoneal reflection) in 11 patients, the lower rectum (below the peritoneal reflection) in 23, and the anal canal in 19. Abdominoperineal resections were performed in 31 patients, anterior resections in 15, Hartmann’s procedure in 2, and total pelvic exenteration in 5. Ten patients underwent combined resection of anterior organs, and one patient underwent resection of the sacral bone. Among these 53 patients, 9 had distant metastases, 3 had liver metastases limited to one lobe, 3 had dissemination to the adjacent peritoneum, 2 had paraaortic nodal involvement, and 1 had lung metastasis limited to one lobe, all of which could be removed by a combined resection at the primary operation.

No patients received preoperative adjuvant therapy. As postoperative adjuvant chemotherapy, 48 patients received oral administration of anticancer chemotherapeutic agents (5-fluorouracil, 5′-DFUR, HCFU, or UFT) (2–60 months; central period 14 months). Among these, 16 patients received further intravenous chemotherapy of mitomycin C (12 patients; total dose 10–30 mg, average 16.7 mg) and/or 5-fluorouracil (6 patients, total dose 2.0–32.0 g, average 20.7 g), and 2 patients received intraarterial hepatic artery infusion of 5-flurouracil after hepatectomy (total dose 2.3 and 3.3 g). Five patients received no adjuvant chemotherapy. Three of eight patients who were found to have circumferential surgical margin involvement after microscopic examination received postoperative radiotherapy (45–50 Gy).

There were no postoperative deaths. All patients were followed up for more than 3 years or until death; the average follow-up period for the survivors was 71 months (range 36–126).

Division of Lateral Area and Surgical Procedure

The lateral area was defined and divided into the following eight regions based on the Japanese Classification of Colorectal Carcinoma: the middle rectal root (outside pelvic plexus), internal iliac, common iliac, external iliac, obturator, median sacral, aortic bifurcation, and inguinal regions (Fig. 1). ¹⁶

Figure 1. The eight regions of the lateral area.

Lateral node dissection was performed after en bloc resection of the rectum, the mesorectum with visceral pelvic fascia, and the lateral sidewall structures including the pelvic nerve plexuses. Total mesorectal excision was performed in all patients with lower tumors below the peritoneal reflection, and at least 4 cm of the anal mesorectum was dissected in those with tumors above the peritoneal reflection. Lateral dissection was performed in each lateral division by using a sharp incision technique under direct vision. Nerve plexuses were not spared, with the intention of performing complete dissection of the middle rectal root region. Groin dissection was added when preoperative biopsy of swollen inguinal nodes revealed cancer involvement. After excision, lymph nodes were removed from the dissected specimens and examined pathologically by each lateral region.

Data Analysis

Survival differences between patients were examined by the Kaplan-Meier method with log-rank analysis. Only deaths from recurrent carcinoma were counted as events, and noncancer deaths were censored to the date of the last follow-up. Univariate analyses were performed using survival and locoregional recurrence as the outcome variables. Locoregional recurrence was defined as a recurrence occurring at the intrapelvic (except for anastomosis), perineum, or inguinal regions. Factors that influenced the survival of patients by univariate analysis to a level of P < .1 were entered into the stepwise forward Cox regression model using StatView 4.11 software (Abacus Concepts, Inc., Berkeley, CA) to determine which factors independently influenced patient survival.

RESULTS

Among the lateral lymph node regions, the middle rectal root nodes showed the highest incidence of nodal involvement, followed by the internal iliac and obturator nodes (Table 1).

Table 1. INCIDENCE OF NODAL INVOLVEMENT BY THE LATERAL DIVISION

* Lateral nodal involvement in only one division.

‡ Lateral nodal involvement in two or more divisions.

The 5-year survival rate among the 53 patients who underwent lateral dissection was 26.6%. Patients without distant metastases had a 5-year survival rate of 32%, whereas no patient with distant metastases survived for more than 3 years after surgery despite lateral dissection (Fig. 2).

Figure 2. Survival comparison of patients with lateral metastasis based on the distant metastases (P < .0001, log-rank test). Present: patients with distant metastases. Absent: patients without distant metastases.

In patients without distant metastases, adjuvant chemotherapy (none, 5 patients; oral agents, 27 patients; and oral agents plus intravenous chemotherapy, 12 patients) had no impact on postoperative survival. Other clinicopathologic factors affecting the survival of patients without distant metastases are detailed in Table 2. Date of operation, patient gender, tumor location (site of tumor anal edge), maximum diameter of tumor, circumferential degree of tumor occupation, and invasion of other organs had no significant impact on disease-specific survival. Age, tumor differentiation, and venous invasion showed a trend to affect the prognostic outcome.

Table 2. CLINICOPATHOLOGIC VARIABLES AND CANCER-SPECIFIC POSTOPERATIVE SURVIVAL OF PATIENTS WITHOUT DISTANT METASTASES

* Cox proportional hazard model (HR, hazard ratio; CI, confidence interval).

‡ Degree of each vessel invasion was assessed according to the Japanese Classification of Colorectal Carcinoma. ¹⁶

Eight of the 44 patients (18%) without distant metastases had histologic circumferential surgical margin involvement, which had a significant impact on prognosis (Fig. 3). Seven of these eight patients showed locoregional recurrence, and all eight died of cancer within 3 years. On the other hand, the 5-year survival rate of patients without circumferential surgical margin involvement reached 39%.

Figure 3. Survival comparison of patients without distant metastases based on the circumferential surgical margin involvement (P < .0001, log-rank test). Positive: patients with surgical margin involvement. Negative: patients without surgical margin involvement.

The region of lateral involvement did not affect survival; patients with cancer involvement of the “outer” lateral region (i.e., the obturator, external iliac, and inguinal regions) did not show an unfavorable survival rate compared with patients with involvement of other lymph node groups (i.e., the middle rectal root, internal iliac, common iliac, median sacral, and aortic bifurcation regions) (Fig. 4). However, the extent of lymph node metastases (mesorectal, lateral, and overall) had a significant impact on prognosis in patients with lateral nodal involvement (Fig. 5), as did the degree of lymphatic invasion. Based on the univariate analyses using the Cox proportional hazard model, the number of metastatically involved nodes in the mesorectum and lateral node regions had a higher hazard ratio (7.0) than that in the mesorectum (2.9) or in the lateral region (2.8). Multivariate stepwise regression analysis revealed that age, circumferential surgical margin involvement, and the overall number of involved nodes had an independent predictive value in determining prognosis (Table 3).

Figure 4. Comparison of survival curves of patients without distant metastases based on the nodal involvement of outer lateral regions (no significance, log-rank test). Outer lateral regions: obturator node, external iliac node, and inguinal node.

Figure 5. Survival comparison of patients without distant metastases based on the extent of lymph node involvement. Mesorectum, mesorectal nodal involvement (P = .0031, log-rank test); lateral, lateral nodal involvement (P = .0111); overall, total number of nodes involved in the mesorectum and lateral region (P < .0001).

Table 3. MULTIVARIATE REGRESSION ANALYSIS FOR SURVIVAL IN PATIENTS WITH LATERAL NODAL INVOLVEMENT

* Degree of each vessel invasion was assessed according to the Japanese Classification of Colorectal Carcinoma. ¹⁶

‡ Only variables found to be significant in the model using a stepwise forward Cox regression model.

Of the 44 patients without distant metastases, locoregional recurrence was observed in 25 (56.8%): intrapelvic recurrence alone in 8, perineal recurrence alone in 2, inguinal recurrence alone in 4, and multiple regions recurrence in 11 patients. Patients with involvement of four or more nodes overall showed a significantly higher incidence of combined recurrence of locoregional and distant organs (Fig. 6). Among the seven patients who showed both involvement of three or fewer nodes overall and locoregional recurrence, three (42.9%) underwent surgery (groin dissection in one and resection of the perineum in two) (Table 4). At the time of writing (more than 3 years after reoperation), all three were alive without disease (follow-up period 41, 51, and 55 months). Of the 18 patients who showed involvement of more than three nodes and locoregional recurrence, four (22.2%) underwent surgical treatment for recurrent disease (total pelvic exenteration in two, groin dissection in one, and resection of the perineum in one), but all four died of disease within a short period (5–39 months).

Figure 6. Recurrent mode with respect to the number of overall (mesorectum + lateral) nodal involvement (*P < .05, chi-square test).

Table 4. PROGNOSIS OF PATIENTS WITH LATERAL NODAL INVOLVEMENT WHO UNDERWENT SURGERY FOR LOCOREGIONAL RECURRENCE

DFI, disease-free interval.

* Time (months) after surgery for first recurrence.

DISCUSSION

Heald and Ryall ¹ first proposed a total mesorectal excision characterized by complete excision of the visceral rectal mesentery, and studies showed that this technique had a lower rate of local failure than the conventional techniques. ^17–19 Currently, this procedure may be the best treatment for patients whose disease is confined to the mesorectum. However, for the more advanced locoregional control of rectal cancers, we cannot disregard the existence of extramesorectal cancer spread. It is essential to know how these diseases are to be treated and which patients may benefit from intensive surgical treatment.

Although the lateral efferent lymphatic drainage system of primary rectal cancer has been recognized since as early as the 1930s, ^20–23 there is no evidence of prognostic improvement by any alternative therapies for such extensive disease other than surgical treatment. Although lateral pelvic dissection has been performed since the early 1950s, ^24–26 it has not been held in high regard in Western countries. The negative feelings about this technique may be based on the results of studies reporting a lack of significant prognostic improvement between patients undergoing and those not undergoing lateral dissection, 5,7,27,28 and it is often concluded that lateral metastases predominantly reflect systemic spread rather than regional disease. ¹⁷ Although the question of whether lateral involvement means regional or systemic disease should be answered only by estimating the prognoses of patients with actual lateral involvement who received thorough lateral dissection, few reports have addressed this matter. To date, the studies that have reported a disappointing outcome after lateral dissection have tended to examine populations that included only a few patients with actual lateral involvement (Table 5). However, based on our survival data and other reports that include a relatively larger number of lateral-positive patients, ^11,12 lateral nodal involvement can be safely said to be a regional disease and to be curable by dissection in about 40% of patients who have no other distant disease.

Table 5. REPORTS ON LATERAL NODAL DISSECTION IN TERMS OF THE NUMBER AND PROGNOSIS OF PATIENTS WITH LATERAL NODAL INVOLVEMENT

* In patients with lower rectal cancer.

† In patients with neither distant metastases nor circumferential surgical margin involvement.

The question of the appropriateness of surgical intervention against extramesenteric disease is a highly important one, because such intervention both provides a prognostic advantage and increases the risk of postoperative problems (i.e., urinary and sexual dysfunction). Enker ²⁹ stated that only one third of patients with mesorectal lymph node involvement also had lateral node involvement, and thus two thirds of patients with mesorectal involvement would not benefit from lateral node dissection. It is difficult to evaluate the effectiveness of dissection against occult metastases to the lateral region, which may explain why lateral dissection has been reported to improve the survival outcome even in Dukes’ B cases. ³⁰ Calculating simply from the proved lateral involvement rate (21%) and the actual 5-year survival rate (27%), the percentage of patients who may derive a survival benefit from surgery was approximately 5% to 6% (21% × 27%) in this study. When performed by surgeons with a thorough knowledge of pelvic sidewall anatomy, the lateral dissection procedure itself should never be related to an increase in the postoperative death rate. ^11,25 However, Hojo et al ¹⁰ reported that lateral node dissection increased the incidence of urine-voiding failure by 31% and sexual impotency by 39% compared with the conventional procedure. Considering such complications after this procedure, more appropriate patient selection criteria would be required to add significant value to this surgical procedure.

Perhaps the most important finding of this study was that the existence of other distant metastases and the total number of involved lymph nodes should be the two keys in choosing a treatment strategy against extramesenteric disease. As to the involved region, involvement of the “outer” lateral regions (inguinal, obturator, and external iliac regions) has been regarded as meaningless to dissect ³¹; however, our results showed that there was no survival difference between patients with and without nodal involvement in the outer lateral regions. However, patients whose total (mesorectal plus lateral) nodal involvement was fewer than four nodes represented one third of lateral node-positive patients without distant metastases, and the 5-year survival rate of such patients reached 75% in our 10-year series. Conversely, patients with involvement of four or more nodes showed a significantly high incidence of combined recurrence of locoregional and distant organs. These results show the possibility of effective case selection; moreover, they underscore the importance of precise diagnosis with regard to the nodal state. ³²

With regard to surgical technique, this study left the door open to further trials to improve the survival and functional outcome. Eight patients, as many as 18% of patients without distant metastases, were histologically revealed to have unexpected circumferential surgical margin involvement, which was closely correlated with both locoregional recurrence and poor prognosis, as previously reported in patients without lateral involvement. ^3,33,34 At the same time, the presence of invasion to other organs, once it was thoroughly resected, did not affect survival. These results might indicate that obtaining adequate surgical margins by aggressive, wide resection (synchronous resection of the organ adjacent to the invasive front) prolongs the survival of patients with lateral involvement. Moriya et al ¹¹ showed the effectiveness of en bloc excision of iliac vessels by showing the long-term disease-free survival of patients with even eight or more involved nodes. In addition, favorable functional outcomes after nerve-preserving surgery combined with lateral dissection have been reported by several surgeons. ^12,35–39

Some believe that the prognostic outcomes after lateral dissection should be based on randomized prospective studies. ^14,15,40 However, considering the prognostic outcome obtained by lateral dissection in this study, we cannot accept from an ethical point of view a randomized trial, in which some patients having a high incidence of lateral lymph node involvement would be assigned to a group without any regional therapy (i.e., lateral node dissection or radiation therapy). Recognizing the benefit of lateral dissection, it is important for us to concentrate on establishing a treatment strategy for patients with such unfavorable variables as distant metastases and massive nodal involvement, as well as on verifying the appropriateness of nerve-preserving surgery for patients who need lateral dissection.

Acknowledgment

The authors thank Dr. John Murphy (Cancer Research Center, Mercy Hospital, Cork, Ireland) for reviewing the manuscript.