Perineural invasion in pT3N0 rectal cancer

The incidence and its prognostic effect

Abstract

BACKGROUND:

The authors' purpose was to explore the incidence and prognostic significance of perineural invasion (PNI) in pT3N0 rectal cancer.

METHODS:

Pathologic materials from resected specimens of 173 patients with pT3N0 rectal cancer were retrospectively collected. PNI-positivity was categorized into 2 groups: surrounding the nerve sheath (SS-PNI) and invading through the nerve sheath (TS-PNI). The rate of PNI-positivity was compared with PNI as initially recorded in the original reports. Patients' outcome was studied in groups with different PNI status, and multivariate analysis was performed to determine its prognostic value.

RESULTS:

In this retrospective analysis, PNI-positivity was found in 24.3% of all cases, in which SS-PNI and TS-PNI were 11% and 13.3%, respectively, and was related to lymphovascular invasion. Only 7.5% of patients' specimens were reported as PNI-positive in the original reports. Detection of SS-PNI was likelier to be missed than TS-PNI. The rates of local recurrence, disease-free survival, and overall survival at 5 years were similar between the groups of SS-PNI and TS-PNI. The 5-year local recurrence rate was more than 2.5-fold higher in the PNI-positive group compared with the PNI-negative group (22.7% vs 7.9%, respectively; P = .017). Multivariate analysis proved that PNI-positivity was the only independent risk factor for predicting 5-year local recurrence rate, whereas only sampled lymph nodes was related to 5-year disease-free survival and overall survival.

CONCLUSIONS:

PNI is a common pathologic feature in rectal cancer. The definition of PNI should include SS-PNI and TS-PNI. Rectal cancer patients who are PNI-positive are at higher risk of local recurrence and should be considered for more intensive treatment. Cancer 2011. © 2010 American Cancer Society.

Colorectal cancer is the fifth most common malignancy in China. In Shanghai, one of the most developed areas of China, colorectal cancer has become the second most frequently diagnosed cancer. Among 6125 cases of colorectal cancer diagnosed in Shanghai in 2006, nearly 40% (2487 cases) of patients had rectal cancers.1 Determining the optimal treatment of rectal cancer is a complex process. Except for the risk of distant metastasis, local recurrence is an important issue in patients with rectal cancer. Although combined chemoradiotherapy has become the standard treatment for rectal cancer, the magnitude of the improvement in local control with the addition of radiation after total mesorectal excision (TME) is limited, and the effect of the addition of radiation on patients' survival is controversial.2, 3 For pT3N0 rectal cancer, adjuvant or neoadjuvant chemoradiotherapy appears to be an excessive treatment in some studies.4, 5 Identifying high-risk factors of recurrence may be more valuable to optimize the treatment in patients with pT3N0 rectal cancer.

Perineural invasion (PNI) is one of the most widely studied pathologic factors in malignant tumors. As the peripheral nerve fiber is embraced by nerve sheath, PNI was defined by Batsakis as tumor cell invasion in, around, and through the nerves in neurotropic carcinomas.6 However, this definition was challenged by some authors.7 The prognostic significance of PNI is widely advocated in malignancies of the head and neck and prostate,8-10 but controversies exist in colorectal cancer.11-14 In T1-2 early rectal cancer, PNI is one of the adverse pathologic features in selecting patients undergoing local excision. But whether it can be used as a criterion to select patients receiving chemoradiotherapy in T3N0 patients is still unknown to us.

In this study, by rereviewing the PNI status in patients with pT3N0 rectal cancer, the goal was to estimate the true incidence of PNI in rectal cancer and to explore the prognostic significance in pT3N0 rectal cancers. As PNI is routinely reported in pathologic reports at our institution, the rereviewed data were also cross-compared with the original reports.

MATERIALS AND METHODS

Patients and Database

The American Joint Committee on Cancer and International Union Against Cancer (AJCC/UICC) T3N0M0 patients with rectal cancer, whose tumors were located within 12 cm from anal verge, were collected for the purpose of this study. Between January 1996 and December 2004, a total of 221 T3N0M0 patients with rectal cancer were collected according to the colorectal database at Fudan University Shanghai Cancer Center, Shanghai, China. All patients underwent anterior resection or abdominoperineal resection for curative intent, and the operations were performed under the principle of total mesorectal excision for middle and low rectal cancers. Before 2005, the standardized neoadjuvant radiotherapy or chemoradiotherapy for rectal cancer was not well established in this hospital; none of the patients received adjuvant or neoadjuvant radiotherapy or chemoradiotherapy. Adjuvant chemotherapy alone with 5-Fu–based regimens without addition of radiotherapy was used in selected patients. Forty-eight patients were excluded: 23 patients (10.4%) were lost at the beginning of follow-up, and for the other 25 (11.3%), the hematoxylin and eosin (H&E)-stained slides or formalin-fixed and paraffin-embedded (FFPE) blocks were unavailable. Finally, a total of 173 patients with T3N0M0 rectal cancer were included in the study.

All patients were asked to follow up every 3 to 6 months after treatment in the first 3 years and every 6 to 12 months thereafter. Digital examination and serum carcinoembryonic antigen (CEA) tests were performed at each session. Radiologic study and colonoscopy were performed every 6 to 12 months or when clinically indicated. All surviving patients were followed up between January and July 2009 in addition to their scheduled follow-up by registered mail, telephone, or local cancer registration system.

Histologic Analysis

All of the H&E slides of the 173 T3N0 patients were rereviewed by 2 pathologists with a special interest in colorectal cancer who were blinded to the patients' clinical information. Similar to the study of Liebig et al,14 PNI-positive was defined as tumor in the perineural space that surrounded at least one-third of the nerve without invading through the layer of epineurium (surround-sheath PNI [SS-PNI], Fig. 1A) or tumor cells within any of the 3 layers (epineurium, perineurium, or endoneurium) of the nerve sheath (through-sheath PNI [TS-PNI], Fig. 1B). In our study, PNI-positive were classified into 2 subgroups (SS-PNI and TS-PNI) and analyzed separately. When both SS-PNI and TS-PNI were presented in reviewed slides, they were also considered TS-PNI in the current study. Each H&E slide of the tumor represents one tumor block sampled from the original primary lesion. The number of reviewed slides was also recorded to evaluate the relationship between the PNI detection rate and number of examined slides. Other pathologic variables, such as histopathology, lymphovascular invasion (LVI), tumor grade, invasion depth, and circumferential margin (CRM) were also reviewed. The CRM, which was defined as the deepest penetration of the tumor and the edge of resected soft tissue around the rectum, was categorized into three groups: <1 mm, 1-2 mm, and >2mm.

Figure 1. Perineural invasion in T3N0 rectal cancer (A) The cancer cells surround the neural sheath without invading through the sheath (arrows). (B) The cancer cells invade through the neural sheath (arrows).

Statistics

Local recurrence (LR) time was calculated from the completion of surgery to the time when cancer recurred in the surgical bed or pelvis. Disease-free survival (DFS) was calculated from the completion of surgery to the time of tumor recurring locally or distantly. The correlation of PNI with other pathologic characteristics was analyzed by chi-square test. Five-year LR, DFS, and overall survival (OS) rates were calculated using the Kaplan-Meier method, and the log-rank test was used to compare variables in the univariate analysis. Multivariate analysis was performed with Cox regression, and hazard ratios were calculated including the 95% confidence interval. P < .05 was considered statistically significant.

RESULTS

Patients' Characteristics and Outcome

The patients' clinicopathologic characteristics are listed in Table 1. The median age in our series was 57 years (range, 24-88 years). High rectal lesions, whose tumors (distal edge) were located above the peritoneal reflex, were found in 18.5% of patients. A total of 98 patients (56.6%) underwent adjuvant chemotherapy with the 5-Fu based regimens.

Table 1. Patients' Clinicopathologic Characteristics and Outcomes of the 173 Patients With T3N0 Rectal Cancer

	Patients (n=173)		Local Recurrence		Disease-Free Survival		Overall Survival
	No.	%	5-Year %	P	5-Year %	P	5-Year %	P
AR indicates anterior resection; APR, abdominoperineal resection; AC, adenocarcinoma; MSC, mucinous adenocarcinoma or signet ring cancer; CRM, circumferential margin; SS-PNI, surround-sheath perineural invasion; TS-PNI, through-sheath perineural invasion.
Sex
Male	109	63	11.7	.92	75.1	.776	79.7	.469
Female	64	37	10.6		71.1		75.3
Age, y
≤60	100	57.8	8.0	.167	74.5	.733	81.5	.187
>60	73	42.2	15.6		72.0		73.3
Tumor location
Above peritoneal reflex	32	18.5	10.3	.814	81.5	.135	81.4	.374
Below peritoneal reflex	141	81.5	11.5		71.7		77.2
Tumor size
<5 cm	69	39.9	11.5	.810	70.1	.156	79.6	.654
≥5 cm	104	60.1	11.0		75.7		77.1
Operation
AR	99	57.2	9.9	.360	76.1	.160	80.2	.073
APR	74	42.8	13.2		70.1		75.4
Histopathology
AC	150	86.7	11.4	.781	72.5	.560	77.1	.413
MSC	23	13.3	10.4		79.9		84.2
Tumor grade
Medium-low	138	79.8	10.2	.330	73.3	.857	77.7	.796
High	35	20.2	15.6		74.3		79.0
CRM
<1 mm	0	0	—	.90	—	.47	—	.22
1-2 mm	21	12.1	11.5		75.4		77.8
>2 mm	152	87.9	10.6		61.1		68.7
Lymphovascular invasion
Yes	25	14.5	18.6	.220	66.9	.504	63.7	.201
No	148	85.5	10.1		74.6		80.0
Sampled lymph nodes
<12	84	48.6	16.4	.042	62.1	.001	69.8	.025
≥12	89	51.4	6.7		84.2		85.8
Adjuvant treatment
No	75	43.4	13.9	.468	69.9	.703	72.7	.323
Chemotherapy	98	56.6	9.2		76.4		82.5
Perineural invasion
Negative	131	75.7	7.9	.017	75.3	.478	78.3	.545
SS-PNI	19	11.0	22.7		67.3		77.8
TS-PNI	23	13.3

With a median follow-up time of 49 months (range, 9-145 months) for all patients, 36 patients (20.8%) died since last follow-up. Local or distant recurrence was observed in 42 patients (24.3%), in which 13 patients had local recurrence only, 25 had distant metastasis only, and 4 had synchronous local and distant recurrence. Of all 173 patients, the 5-year LR rate, DFS rate, and OS rate were 11.3%, 73.5%, and 77.9%, respectively.

Pathologic Analysis of PNI

PNI-positivity was found in 42 cases (24.3%) when the slides were rereviewed, in which 11.0% were SS-PNI and 13.3% were TS-PNI. Nine cases were found to be both SS-PNI–positive and TS-PNI–positive (5.2%). However, only 7.5% (13 cases) of patients were detected as PNI-positive in the original pathologic reports. Compared with the rereviewed data about PNI, in 131 rereviewed PNI-negative patients, only 2 cases (1.5%) of PNI-negative were misdiagnosed as PNI-positive in the original reports. However, the diagnosis of PNI-positive was missed in 73.8% (31 of 42 cases) of patients according to the original reports. Meanwhile, SS-PNI was found to be more easily neglected by pathologists than TS-PNI (missed diagnosis rate, 89.5% in SS-PNI vs 60.9% in TS-PNI, P = .036; Fig. 2).

Figure 2. The association between missed diagnosis rates is shown according to original reports and the different subgroups of patients who were perineural invasion (PNI)-positive (P = .036).

The incidence of PNI was studied with different pathologic features (tumor location, tumor size, histopathology, tumor grade, T stage, LVI). A median of 4 slides were rereviewed for each case (range, 1-18), and the number of reviewed slides were also analyzed for PNI incidence. The rate of PNI-positivity was found to be significantly higher in patients who were LVI-positive (PNI-positive rate, 52% in LV+ vs 19.6% in LV−; P < .001), whereas the number of reviewed slides was not related to the detection rate of PNI-positive (P = .23). In 42 patients who were PNI-positive, the proportion of SS-PNI and TS-PNI wasn't found to be correlated with other pathologic features.

Prognostic Significance of PNI in LR, DFS, and OS

The prognostic significance of PNI status and other clinicopathologic variables was analyzed. The rates of LR, DFS, and OS at 5 years were compared for each variable in univariate analysis (Table 1). Because the rates of 5-year LR, DFS, and OS were 25.1%, 70.2%, and 69.6% in the group of SS-PNI compared with the rates of 19.7%, 63.8%, and 70.7% in the group of TS-PNI (P = .795,.977, and.456, respectively), the 2 subgroups were combined together in prognostic analysis. Significantly lower 5-year DFS and OS rates were observed in patients with less than 12 sampled lymph nodes (P = .001 and.025, respectively). The 5-year LR rate was more than 2.5-fold higher in the PNI-positive group compared with the PNI-negative group (22.7% vs 7.9%, respectively; P = .017; Fig. 3), whereas no significant difference was found for DFS (67.3% vs 75.3%; P = .478) and OS (70.0% vs 78.3%; P = .545).

Figure 3. Kaplan-Meier survival curves of local recurrence rate are shown for patients who were perineural invasion (PNI)-positive and PNI-negative.

Variables analyzed in univariate analysis were included in multivariate analysis by using Cox regression method. PNI-positivity was found to be the only independent risk factor for predicting LR (hazard ratio = 2.70; 95% CI, 1.03-7.06; P = .043), whereas PNI status was found not to be related to DFS (hazard ratio = 1.04; 95% CI, 0.49-2.18; P = .93) and OS (hazard ratio = 0.98; 95% CI, 0.43-2.25; P = .97). Sampled lymph nodes <12 was found to be the only independent risk factor for predicting DFS (hazard ratio = 2.85; 95% CI, 1.48-5.49; P = .002) and OS (hazard ratio = 2.14; 95% CI, 1.08-4.23; P = .029).

DISCUSSION

Combined-modality therapy is widely advocated for improving the local control rate in patients with clinical-staged T3N0 rectal cancer, as nearly one-quarter of cT3N0 patients turn out to be node-positive even after chemoradiotherapy.5, 15 However, for pathological-staged T3N0 rectal cancer, the value of adjuvant radiotherapy for all patients remains controversial. Park et al did not find a significant difference in LR rate between patients who did and did not receive adjuvant radiotherapy in 390 pT3N0 patients.16 Other studies also reported only a 4% to 8.2% LR rate for pT3N0 patients of low risk with total mesorectal excision alone.17-20 Identifying pT3N0 patients, who are at high risk of local failure, will help optimize the treatment modality and diminish excessive toxicities caused by adjuvant treatment.

PNI refers to cancer involving the space surrounding a nerve, and was first described in a primary head and neck tumor in 1862 by Neumann.21 PNI has been studied in many malignancies, including prostate cancer, head and neck cancer, and cancer in the gastrointestinal tract,9, 22-24 yet there is still no agreement on a clear definition of PNI-positive. A stringent definition of PNI requires that tumor cells be observed inside the perineurial layer,7 but this definition would not include the situation of tumor cells existing within epineurium. Another definition of PNI advocated by many pathologists: tumor cells are observed within any layer of the nerve sheath.6, 25-27 However, this definition would exclude the situation where tumor cells abut the nerve without invading through the nerve sheath, a situation commonly observed in pathology slides.

A broader definition widely used in the literature would be inclusive of both of the circumstances: tumor cells surround at least one-third of the nerve without invading through the nerve sheath (SS-PNI), as well as tumor cells within any of the 3 layers of the nerve sheath (TS-PNI).10, 14, 28, 29 In our study, we used the broad definition of PNI in pT3N0 rectal cancers, but we further studied the 2 tumor-nerve relationships (SS-PNI and TS-PNI) by analyzing their outcomes separately. Patients in the SS-PNI group and in the TS-PNI group both had significantly higher 5-year LR rates than that of the PNI-negative group, and a similar 5-year LR rate was also observed between the SS-PNI group and TS-PNI group. As differences in prognosis are the gold standard for defining a prognostic factor, our results provide clear and direct evidence that both SS-PNI and TS-PNI should be defined as PNI-positive.

In colorectal cancer, the incidence of PNI varies from series to series from <10% to 33%14, 30-32 Liebig et al observed a mean of 22% of PNI in stage I-IV colorectal cancer. However, the detection rate rose from 0.5% to 22% merely by the author rereviewing the slides,14 suggesting that PNI may be missed or ignored by clinical pathologists. Similarly, although PNI is a required component of a rectal cancer pathology report at our institution, nearly 75% of PNI-positive cases were originally missed.

Several reasons may contribute to underreporting. First, as described above, disagreements in the definition of PNI will affect the decision to declare a tumor PNI-positive. In our study, 45% (19 of 42) of PNI-positive cases invaded the nerve sheath only and should have been categorized as SS-PNI only. This tumor-nerve relationship is clearly easier to be missed and underreported compared with TS-PNI. That was the case in this series, with underreporting of 89.5% of SS-PNI compared with 60.9% of TS-PNI. Second, nerve invasion may be difficult to recognize in some circumstances. Inflammatory cells or large, mucinous pools may obscure the presence of tumor cells around nerves,33 and the peripheral nerve may be more difficult to recognize within the main body of the tumor. Careful examination and rereview of slides by pathologists specified in this area may improve the detection rate. By using the same definition of PNI in 40 patients of oral cavity squamous cell carcinoma, Krutz et al reported that the PNI-positive detection rate was increased from 30% in original reports to 62% in the rereview study.34

For all of these technical reasons, the main impediment to proper reporting may be the standards of the College of American Pathologists: PNI status is not a required element in the pathology report for a colorectal cancer, presumably because the prognostic value of PNI has not been established in colorectal cancer.35, 36

This and other reports may lead to a change in this policy. Poorer outcome has been demonstrated by several studies in patients with PNI-positive colorectal cancer. In a series of 269 stage I-IV patients with colorecal cancer, the 5-year DFS and OS was significantly higher in those with PNI-positive tumors, and PNI was proven to be an independent prognostic factor for both cancer-specific overall and disease-free survival.14 In another series of 341 stage II patients, PNI was found to be significantly related to the 5-year DFS rate, not for overall survival rate.25 Shirouzu et al also reported that patients with PNI had a higher LR rate and lower 8-year OS rate in stage III rectal cancer.37 In 622 patients with stage I-III rectal cancer undergoing anterior resection, Law et al found PNI was an independent risk factor for predicting LR and disease-specific survival.23 Similar results were also obtained by other studies.30, 31, 38 However, most of these studies combined colon and rectal cancer or cancer with different stages together. pT3N0 rectal cancer, which was one of the most controversial areas in the management of colorectal cancer, was seldom focused on specifically. And the different criteria of PNI made the results more difficult to explain. In 173 pT3N0 rectal cancers, we found PNI was the only independent risk factor for predicting the 5-year LR rate, whereas the number of sampled lymph nodes was proven to be related to 5-year DFS and OS rates. As radiotherapy has mainly been proven to improve local control, PNI can serve as a good risk factor for selecting patients receiving radiotherapy in pT3N0 rectal cancer.

Because most cT3N0 rectal cancer patients undergo neoadjuvant chemoradiotherapy, the pathologic evaluation of the specimens for this and other factors may be less meaningful. Traditional assessment criteria for primary tumor, lymph nodes, distal or circumferential margin, and PNI or LVI may not be relevant in treated specimens. Our study included specimens without any preoperative treatment, providing the natural feature of PNI in rectal cancer. Otherwise, different from our knowledge in sampled lymph nodes, increasing the number of reviewed slides seems not related to the detection rate of PNI. This result suggests that the presence of PNI may be related to a certain biologic mechanism of the primary tumor. Some authors use nerve-specific staining, such as S-100 protein, to improve the detection rate.34 Some authors also studied the expression of neurotrophins, such as NGF, BDNF, and NT-3, in tumor-invasive behavior.39, 40 Further studies in this area are needed to clarify the association between the pathogenesis of PNI and cancer invasion.

Perineural invasion is a common pathologic feature in rectal cancer. Both tumor cells surrounding at least one-third of the nerve and invading through the nerve sheath should be defined as PNI-positive. For pT3N0 rectal cancer, PNI was an independent risk factor predicting local recurrence.

CONFLICT OF INTEREST DISCLOSURES

The authors made no disclosures.