Ann Surg Oncol DOI 10.1245/s10434-015-4585-1
ORIGINAL ARTICLE – COLORECTAL CANCER
Impact of Total Lymph Node Count on Staging and Survival After Neoadjuvant Chemoradiation Therapy for Rectal Cancer Matthew D. Hall, MD, MBA1, Timothy E. Schultheiss, PhD1, David D. Smith, PhD2, Marwan G. Fakih, MD3, Joseph Kim, MD4, Jeffrey Y. C. Wong, MD1, and Yi-Jen Chen, MD, PhD1 Department of Radiation Oncology, City of Hope National Medical Cancer, Duarte, CA; 2Division of Biostatistics, City of Hope National Medical Cancer, Duarte, CA; 3Department of Medical Oncology and Therapeutics Research, City of Hope National Medical Cancer, Duarte, CA; 4Department of Surgery, City of Hope National Medical Cancer, Duarte, CA 1
ABSTRACT Purpose. Current guidelines recommend that a minimum of 12 lymph nodes (LNs) be dissected to accurately stage rectal cancer patients. Neoadjuvant chemoradiation therapy (CRT) decreases the number of LNs retrieved at surgery. The purpose of this study was to assess the impact of the number of LNs dissected on overall survival (OS) for localized rectal cancer patients treated with neoadjuvant CRT. Methods. Treatment data were obtained on all patients treated for rectal cancer (2000–2013) in the National Oncology Data AllianceTM, a proprietary database of merged tumor registries. Eligible patients were treated with neoadjuvant CRT followed by surgery and had complete data on number of positive LNs, number of LNs examined, and treatment dates (n = 4565). Results. Hazard ratios for OS decreased sequentially with increasing number of LNs examined until a maximum benefit was achieved with examination of eight LNs. On multivariate analysis, age, sex, race, marital status, grade, ypT stage, ypN stage, type of surgery, margin status, presence of pathologically confirmed metastasis at surgery, and number of LNs examined were significant predictors of OS. Conclusions. Examination of eight or more LNs in rectal cancer patients treated with neoadjuvant CRT resulted in accurate staging and assignment into prognostic groups with an ensuing improvement in OS by stage. This study suggests that eight LNs is the threshold for an adequate lymph node dissection after neoadjuvant CRT.
Accurate identification of the extent of disease is essential in the diagnosis and treatment of any malignancy. Prognosis for localized rectal cancer is primarily determined by the pathologic tumor (pT) stage and the number of positive lymph nodes (LNs). Neoadjuvant chemoradiation therapy (CRT) followed by total mesorectal excision (TME) is preferred over TME and adjuvant CRT in T3–T4 and/or node-positive rectal tumors due to tumor downsizing, increased sphincter preservation rates, and reductions in local recurrence and treatment-related complications.1–4 The current National Comprehensive Cancer Network (NCCN) guidelines recommend that a minimum of 12 LNs be dissected to accurately stage and forecast prognosis in patients with rectal cancer.5–9 However, neoadjuvant CRT decreases the number of LNs retrieved at surgery, and large series indicate that fewer than 50 % of patients achieve this level of dissection.10,11 The current guidelines to dissect 12 or more LNs, based on data collected with upfront surgery, have been questioned.12,13 The optimal number of LNs dissected after neoadjuvant CRT remains poorly characterized. The purpose of this study was to assess the impact on overall survival (OS) of the number of LNs dissected in rectal cancer patients treated with neoadjuvant CRT followed by surgery using a large database and to quantify the effects of prognostic factors. METHODS AND MATERIALS Patient Selection
Ó Society of Surgical Oncology 2015 First Received: 3 February 2015 M. D. Hall, MD, MBA e-mail: [email protected]
Data on all patients diagnosed with rectal cancer from 2000 through 2013 were extracted from the National Oncology Data AllianceTM (NODA), a proprietary database of merged tumor registries (Elekta/IMPAC Medical Systems,
M. D. Hall et al.
Inc., Sunnyvale, CA, USA). The NODA captures newly diagnosed cancer cases at more than 150 hospitals in the US and is fully compliant with American College of Surgeons (ACOS) regulatory requirements. Data in the NODA registries contain exactly the same data submitted to state tumor registries and to the Surveillance, Epidemiology, and End Results (SEER) database, in regions that participate in SEER.14 The study population included all patients treated with neoadjuvant CRT followed by surgery, with complete data on the number of LNs dissected, the number of positive LNs, and the treatment start dates with chemotherapy, radiation therapy, and surgery. Patients who underwent shortcourse neoadjuvant radiation therapy, had metastatic disease at diagnosis, or had a history of prior malignancy were excluded, leaving 4565 evaluable patients. Patient characteristics included age, sex, race, marital status, clinical tumor (cT) and nodal (cN) stage, post-therapy pathologic tumor stage (ypT stage), grade, surgical margin status, presence of pathologically confirmed metastatic disease at surgery, year of diagnosis, year of treatment, number of LNs examined, number of positive LNs, type of surgery performed, radiation therapy dose, concurrent chemotherapy agent, and medical comorbidities. This study was conducted with the approval of the Institutional Review Board. The primary objective was to determine the number of examined LNs associated with the lowest mortality risk after trimodality therapy. Participating institutions in the NODA regularly update patient files with vital status for OS and cause-specific survival (CSS). Statistical Analysis Statistical analysis was performed using SPSS version 18.0 (SPSS Inc., Chicago, IL, USA) after patient-identifiers were removed from the data set. Differences between categorical variables were compared using the v2 test, and the t test was used for continuous variables. Multivariate proportional hazards analysis (MVA) with backward stepwise regression was used to identify factors associated with OS.15 Two-sided p values \0.05 were considered statistically significant. The threshold for the number of LNs examined was determined by examining the trend of the hazard ratios (HRs) on MVA and identifying naturallyoccurring breakpoints with similar survival rates. The proportional hazards assumption was tested for a non-zero slope in the generalized linear model using scaled Schoenfeld residuals. Survival curves were plotted using the Kaplan–Meier method, and compared using the logrank test.
RESULTS Patient Characteristics and the Distribution of Lymph Nodes (LNs) Examined Patient demographics and tumor characteristics are presented in Table 1. Median age was 61 years (interquartile range [IQR] 51–68). Based on the results of the threshold analysis, significant differences in clinical characteristics between patients with \8 and C8 LNs examined are also shown. Patients with \8 LNs examined were more likely to be of older age and be treated with abdominoperineal resection (APR) versus low anterior resection (LAR) than patients with C8 LNs dissected. All other comparisons showed no significant differences between the groups, or found more favorable attributes in patients with \8 LNs examined. Median follow-up was 3.5 years for surviving patients. At the close-out date, 26 % of patients had died. The median number of LNs examined was 11 (IQR 6– 16). In total, 53.1 % of patients had \12 LNs dissected. Multivariate Analysis and Threshold Analysis for Number of LNs Examined Multivariate Cox models were fit for OS, and regression diagnostics showed no violations of the proportional hazards assumptions. On MVA, age, sex, race, marital status, grade, ypT stage, post-therapy pathologic nodal stage (ypN stage), type of surgery performed, surgical margin status, presence of pathologically confirmed metastasis at the time of surgery, and number of LNs examined were significant predictors of OS (Table 2). Treatment year, cT stage, and cN stage were not significant. Exclusion of patients with pathologically confirmed metastasis at surgery (108 cases; 2.4 %) caused no significant change in the results, and this parameter was included in the model. Adjusting for the effect of other prognostic factors, number of LNs examined was a significant predictor of OS on MVA (p = 0.008). HRs for OS decreased sequentially with increasing number of LNs examined until a maximum benefit was achieved with examination of eight LNs (Table 2). Figure 1a illustrates that no further reduction in the HRs for OS was observed after eight LNs were examined. To test the validity of this finding, we separately examined the subgroup of patients with zero positive LNs (n = 3160) and only one positive LN (n = 493). MVA was repeated for each subgroup, and the results are shown in Fig. 1b and c, respectively. A similar pattern was observed, indicating that an adequate nodal dissection must contain at least eight LNs.
Lymph Node Count and Survival in Rectal Cancer TABLE 1 Patient demographics in relation to number of lymph nodes examined Characteristics
Total
Lymph nodes retrieved \8
C8
4565
1511
3054
Median
61
62
60
Interquartile range
52–69 53–70
51–68
Male
63.2
62.7
63.5
Female
36.8
37.3
36.5
No. of patients
TABLE 1 continued Characteristics
Total
p Value
Age, years \0.01
Sex
Race 0.46
\8
C8
Negative
91.4
90.7
91.7
Positive
5.9
5.4
6.2
Unknown
2.7
3.9
2.1
97.6
97.6
97.6
2.4
2.4
2.4
5-FU or capecitabine
84.9
84.8
85.0
Other/unspecified
15.1
15.2
15.0
Metastasis found at surgery No Yes
0.62
Lymph nodes retrieved
88.6
88.6
88.6
Black
6.6
6.8
6.5
\45
2.5
3.0
2.3
Asian/American Indian
3.2
3.6
3.0
45.0–50.4
67.0
65.8
67.7
Unknown
1.6
1.0
1.9
Single Married
32.0 60.2
29.0 62.8
33.4 58.9
Unknown
7.8
8.2
7.7
0.01
3.1
4.3
2.5
T2
9.4
10.8
8.7
T3
64.7
58.9
67.5
T4
4.2
4.9
3.9
Unknown
18.6
21.1
17.4
0
47.2
53.5
44.1
1
30.5
23.5
34.1
2
3.6
1.8
4.5
Unknown
18.7
21.2
17.3
1 2
7.1 66.8
6.4 69.3
7.4 65.6
3
10.9
9.4
11.6
4
0.5
0.3
0.7
Unknown
14.7
14.6
14.7
0.96
0.78
Radiation therapy dose (Gy) 0.13
50.41–54.0
12.0
12.0
12.0
[54
4.0
3.8
4.0
Unknown
14.5
15.4
14.0
Cardiovascular disease
25.0
22.4
26.3
Diabetes mellitus
7.0
6.6
7.2
0.42
Respiratory disease
2.8
2.2
3.1
0.08
Medical comorbidities
cT stage T1
0.31
Concurrent chemotherapy
White
Marital status
p Value
\0.01
\0.01
Data are expressed as percentages unless otherwise stated; two-sided p values \0.05 were considered statistically significant cT stage clinical tumor stage, cN stage clinical nodal stage, ypT stage pathologic tumor stage after neoadjuvant therapy, 5-FU 5-fluorouracil
cN stage \0.01
Grade 0.04
ypT stage ypT0
13.6
15.9
12.4
ypT1–2
30.3
32.7
29.0
ypT3
45.6
39.9
48.5
ypT4
3.3
3.0
3.5
Unknown
7.2
8.5
6.6
\0.01
Type of surgery Low anterior resection
43.9
40.3
45.7
Abdominoperineal resection 33.2
35.5
32.0
Unknown
24.2
22.3
Surgical margin
22.9
\0.01
Five-year OS for the entire cohort was 70.2 %, and the median OS was 9.7 years (95 % CI 8.9–10.5 years). For patients with \8 LNs examined, median OS was 9.4 years (95 % CI 8.6–10.2 years), and 11.1 years (95 % CI 9.5– 12.7 years) for patients with C8 LNs examined. On MVA, patients with C8 LNs examined also experienced a significant CSS advantage compared with patients with \8 LNs dissected (HR 0.87; 95 % CI 0.76–0.99, p = 0.04). Survival Outcomes Based on Pathologic Nodal Involvement Compared to patients with zero positive nodes, the HRs for OS increased with increasing number of positive nodes when divided into five nodal tiers as per the current American Joint Committee on Cancer (AJCC) guidelines (Table 2).16 The Kaplan–Meier curves showed a clear and statistically significant difference in OS in patients with 0, 1, 2–3, 4–6, and C 7 positive LNs (p \ 0.0001), as shown in Fig. 2a. This pattern was also observed for CSS (p \ 0.0001), as shown in Fig. 2b. In the subset of patients
M. D. Hall et al.
(a)
TABLE 2 Hazard ratios for overall survival p Value
HR
95 % CI
All Patients 1.3 1.2 1.1
Number of lymph nodes examined Hazard Ratio
1.0
0–1
0.008
: 1.00
2–3
0.832
0.97
0.73–1.28
4–5
0.401
0.89
0.67–1.17
6–7
0.373
0.88
0.68–1.15
8
0.029
0.69
0.50–0.96
0.4
9–10
0.020
0.71
0.53–0.95
0.3
11–12
0.023
0.71
0.53–0.96
13–14
0.016
0.69
0.52–0.94
C15
0.005
0.69
0.54–0.89
\0.001
1.03
1.02–1.04
0.7 0.6
\0.001
–
–
1.1
0.73
0.64–0.83
1.0
9 ≥1
-1 2 13 -1 4 15 -1 6 17 -1 8
11
8 910
6
7
4
0.8
–
Black
\0.001
1.59
1.28–1.97
American Indian
0.051
1.56
1.00–2.44
0.5
Asian
0.994
1.00
0.62–1.62
0.4
0.7 0.6
0.3
Marital status
1.15
0.89–1.47
ypT3
\0.001
1.86
1.47–2.36
0.2
ypT4
\0.001
2.71
1.93–3.81
0.0
\0.001
–
–
9-
Number of Lymph Nodes Examined
(c)
One Positive Lymph Nodes 1.4
1.0 0.8 0.6
\0.001
1.44
1.20–1.73
\0.001
1.82
1.52–2.18
ypN2a
\0.001
2.25
1.80–2.81
ypN2b
\0.001
3.39
2.70–4.25
Type of surgery – \0.001
–
–
1.28
1.12–1.47
8 ≥1 9
5
-1 16
2
-1
-1 10
ypN1a ypN1b
13
9 8-
7
6
0.4
2
Hazard Ratio
1.2
1-
ypN stage
9
0.293
8
–
\0.001
≥1
–
ypT stage
6
1.02–3.93
-1
2.00
17
0.044
4
4
-1
1.17–2.09
-1
0.92–1.54
1.57
15
1.19
0.002
13
0.196
3
2
2
-1
–
11
–
8
0.001
10
1
7
0.73–0.94
6
0.83
5
0.005
4
Married Grade
3
–
2
–
0
–
1
0.2
Unmarried
Abdominoperineal resection
5
0.9
–
Low anterior resection
Zero Positive Lymph Nodes
1.2
–
\0.001
ypN0
3
(b)
White
ypT1–T2
2
Number of Lymph Nodes Examined
1.3
Race
ypT0
1
0
0.2
Hazard Ratio
Female
0.8
0.5
Sex Male
0.9
35
Age (Continuous)
–
Number of Lymph Nodes Examined
FIG. 1 Hazard ratios for overall survival by number of lymph nodes examined on multivariate analysis. Adjusting for the effect of other factors, hazard ratios for overall survival declined with an increasing number of lymph nodes dissected in (a) all patients, (b) patients with negative lymph nodes, and (c) patients with only one positive lymph node. No further reduction in the hazard ratio for overall survival was observed after eight lymph nodes were examined. The error bars indicate the standard error of the hazard ratio
Surgical margin Negative Positive
– \0.001
–
–
1.62
1.33–1.98
–
–
3.03
2.37–3.89
Metastasis found at surgery No Yes
– \0.001
HR hazard ratio, CI confidence interval, ypT stage pathologic tumor stage after neoadjuvant therapy, ypN stage pathologic nodal stage after neoadjuvant therapy
with C8 LNs dissected, MVA again demonstrated that the number of positive LNs was significant with increasing HRs for OS in patients with 0, 1, 2–3, 4–6, and C7 positive nodes (p \ 0.001). As an exploratory analysis, MVA was performed with number of positive LNs included as a categorical variable to examine the discrimination utility of the current AJCC
Lymph Node Count and Survival in Rectal Cancer All Patients
1.0
Overall Survival
0.8
AJCC N stage N0 N1a N1b N2a N2b
0.6
0.4
4.0
Hazard Ratio
(a)
3.0
2.0
0.2
1.0
0.0 0
Number of Positive Lymph Nodes
Number at Risk 0 LNs Positive
3,160
2,719
2,234
1,774
1,372
1,019
746
1 LN Positive
493
426
342
273
211
158
108
2-3 LNs Positive
457
390
309
225
170
116
83
4-6 LNs Positive
252
207
162
114
67
44
35
≥7 LNs Positive
203
168
116
77
53
33
17
Cause-Specific Survival
(b)
≥8
6
7
5
6
4
5
3
Time (Years)
4
2
3
1
2
0
1
0.0
FIG. 3 Hazard ratios for overall survival by number of positive lymph nodes on multivariate analysis. Groups with similar hazard ratios were combined, yielding five nodal groupings (0, 1, 2–4, 5–7, and C8 positive lymph nodes) that best forecast prognosis in this cohort. The error bars indicate the standard error of the hazard ratio
1.0
0.8
AJCC N stage N0 N1a N1b N2a N2b
0.6
0.4
no additional prognostic information than ypN stage if C8 nodes were examined; ypN stage remained the more useful prognostic variable if \8 nodes were retrieved. DISCUSSION
0.2
0.0 0
1
2
3
4
5
6
Time (Years) Number at Risk 0 LNs Positive
3,110
2,631
2,085
1,600
1,217
891
650
1 LN Positive
488
407
304
227
173
125
93
2-3 LNs Positive
451
365
260
187
137
90
66
4-6 LNs Positive
252
195
137
85
53
33
28
≥7 LNs Positive
199
148
85
49
31
19
10
FIG. 2 (a) Overall survival and (b) cause-specific survival by number of positive lymph nodes. The Kaplan–Meier curves showed a clear and statistically significant association between number of positive lymph nodes and improved overall survival (p \ 0.0001) and cause-specific survival (p \ 0.0001). AJCC American Joint Committee on Cancer, LNs lymph nodes
nodal groupings. Adjusting for the effect of other significant variates, Fig. 3 shows that the HRs for OS increased monotonically in patients with 0, 1, 2–4, 5–7, and C8 positive nodes. When adjoining groups with similar survival rates were merged, Kaplan–Meier analysis again demonstrated a clear and statistically significant difference in OS (p \ 0.0001). For ypN staging, this nodal grouping showed higher discrimination utility as measured by the v2 statistic compared with the AJCC system in this dataset. LN ratio, with patients divided into three tiers (0, 0–30, and [30 %), was significant on MVA when ypN stage was not included in the model. When ypN stage was included, LN ratio was not significant (p * 0.3). LN ratio provided
An adequate LN dissection is critical to delineate the true stage of disease. Published guidelines recommend that a minimum of 12 LNs be dissected to accurately forecast prognosis in patients with colorectal cancer, but did not consider the effects of neoadjuvant therapy.5–7 In rectal cancer, neoadjuvant CRT is employed in appropriately selected patients to improve locoregional control and functional outcomes,1–4 but also significantly reduces the number of LNs retrieved compared to those treated with surgery alone.10,11,17–20 The optimal number of LNs needed to accurately stage and forecast prognosis in patients treated with neoadjuvant CRT has not been previously characterized. In a large patient cohort, our study demonstrated that the number of LNs examined after neoadjuvant CRT was a significant prognostic factor and that OS improved with increasing number of nodes examined until a maximum benefit was achieved with examination of eight LNs. This pattern was also observed on subset analysis in patients with zero positive LNs and only one positive LN. As a result, we recommend that eight or more LNs be dissected to improve the accuracy of nodal staging. Although a target for the minimum number of examined LNs after neoadjuvant CRT was identified, it is important to note that many patients with rectal cancer fail to achieve this level of nodal dissection. In our cohort, 33.1 % of patients had \8 LNs examined. Only 46.9 % met the current guidelines of C12 LNs examined, which is
M. D. Hall et al.
consistent with published multi-institutional and population-based studies.10,18,19,21 In addition to neoadjuvant therapy, patient and disease-specific factors, including age, sex, body mass index (BMI), tumor location, and pT stage, have been reported to impact the number of LNs collected after pelvic surgery.18–23 TME versus proctectomy alone, as well as the experience of the surgeons and pathologists involved, have also been shown to influence the LN yield.22 Even so, TME remains a technically challenging surgery and up to 25 % of cases have an incomplete resection of the mesorectum, even in the hands of experts.24–26 The reason for the pervasiveness of inadequate nodal dissections in rectal cancer is multifactorial and this trend is unlikely to be reversed in the near-term. However, knowledge regarding the number of LNs needed to adequately stage patients treated with neoadjuvant CRT is an important first step. In our study, a five-tier nodal division was found to best forecast prognosis based on number of positive LNs identified after neoadjuvant CRT, consistent with the current AJCC guidelines. In our analysis, the HRs for OS increased monotonically in patients with 0, 1, 2–4, 5–7, and C8 positive nodes. These groupings were determined by examining the HRs for number of positive LNs on MVA (Fig. 3) and merging adjoining groups with similar survival rates. The resulting five-tier system marginally diverges from the AJCC 7th edition nodal staging (0, 1, 2–3, 4–6, and C7 positive LNs). In our cohort, the five-tier nodal grouping we identified had superior discrimination utility and better predicted OS, based on the number of positive LNs, than the current AJCC system. Five-year OS for our study was 77 % for patients with negative LNs, 67 and 58 % for patients with 1 and 2–4 positive LNs, respectively, and 38 and 30 % for patients with 5–7 and C8 positive LNs, respectively. We recommend that this new grouping be considered for the next version of the AJCC pathologic nodal (pN) strata for rectal cancer. This study suggests that the number of LNs examined is an important prognostic parameter in rectal cancer patients treated with neoadjuvant CRT. Prior studies have not demonstrated an association between the number of LNs examined and OS after neoadjuvant CRT.12,27–29 Although single institution series from tertiary care centers may possess more uniform surgical and pathologic technique, sample size may limit their ability to detect smaller effect sizes due to reduced statistical power. Larger sample size is an advantage of pooled registries and population-based databases. Several limitations in our study must be acknowledged. Although this study is the largest to address the impact of LN involvement in patients treated with neoadjuvant CRT followed by surgery for rectal cancer, the data do not enable the opportunity to examine several important factors,
including surgical technique, use of and adherence to adjuvant chemotherapy, and postoperative complications. In addition, the observed detriment in OS due to APR may be related to correlated variables that could not be analyzed, including proximal versus distal tumor location. The outcome measures were limited to OS and CSS, and data regarding local and distant failure cannot be extracted. Patient selection can influence the results in populationbased observational series and must be considered when applying our results to clinical practice. However, the completeness of the available data helps to counter the possibility that the observed LN threshold is due to selection bias. First, the analysis included many known variables that can influence OS in rectal cancer patients, including surgical margin status and even the presence of biopsy-proven metastatic disease at surgery, which are not available in the SEER database. Second, patient demographics and reported medical comorbidities in the subgroup of patients with \8 and C8 LNs dissected were similar, or even favored the group with fewer LNs dissected (Table 1). This suggests that the observed benefit afforded by an ‘adequate’ LN dissection was not the product of surgery provided to inherently healthier individuals. Finally, we acknowledge that observational studies cannot replace randomized data as the standard for outcomes research. However, population-based analyses can provide measures of the effectiveness of therapies in the general population and address important clinical questions that have not been adequately evaluated in randomized trials. Stage migration, due to inappropriate understaging of disease, is an important mechanism to explain the difference in OS observed. Given that the number of positive LNs identified is fundamentally related to the number of LNs examined, a fraction of patients classified as N0 will be understaged.30 This effect can be mitigated by examining a larger number of LNs, specifically C8 in this analysis. In addition, alternative conventions, including LN ratio and the logarithmic odds approach, have been proposed to improve on the existing AJCC staging criteria.31–34 Despite the large body of literature on this topic, as well as our findings, several important questions remain. In this study, cT and cN stage were not significant on MVA after the effect of other variates were included in the model. Earlier work has suggested that the response to neoadjuvant therapy for pT stage and LN yield may not be highly correlated.20,35 Tumor regression is an important metric, and the impact of neoadjuvant CRT on subsequent pT stage and nodal positivity is an important outcome to be further explored. The identification of patients that may be candidates for transanal excision based on objective tumor response criteria, especially in less than optimal surgical candidates, is also of great interest. Finally, in our study,
Lymph Node Count and Survival in Rectal Cancer
the difference in OS for Black and American Indian patients relative to White patients has previously been reported in rectal and other gastrointestinal malignancies. Socioeconomic factors, access to care, and surgical technique have been posited as potential explanations for observed racial disparities but have not been supported by the data.36,37 Racial disparities may also be a function of differing tumor biology, and the molecular and genetic differences in rectal cancer also warrant further investigation. CONCLUSIONS In a large, population-based cohort, our study demonstrates that examination of eight or more LNs in rectal cancer patients treated with neoadjuvant CRT effected accurate staging and allocation into prognostic groups with a resulting improvement in OS. This suggests that eight LNs is the threshold for an adequate nodal dissection after neoadjuvant CRT. A five-tier nodal grouping was found to best forecast prognosis based on the number of positive LNs identified after neoadjuvant CRT. This grouping (0, 1, 2–4, 5–7, and C8 positive LNs) should be considered in the next version of the AJCC pN staging for rectal cancer. DISCLOSURE
None.
REFERENCES 1. Sauer R, Becker H, Hohenberger W, et al. Preoperative versus postoperative chemoradiotherapy for rectal cancer. N Engl J Med. 2004;351(17):1731–40. 2. Peeters KC, Marijnen CA, Nagtegaal ID, et al. The TME trial after a median follow-up of 6 years: increased local control but no survival benefit in irradiated patients with resectable rectal carcinoma. Ann Surg. 2007;246(5):693–701. 3. Ortholan C, Francois E, Thomas O, Benchimol D, Baulieux J, Bosset JF, et al. Role of radiotherapy with surgery for T3 and resectable T4 rectal cancer: evidence from randomized trials. Dis Colon Rectum. 2006;49(3):302–10. 4. Kim DW, Lim SB, Kim DY, et al. Pre-operative chemo-radiotherapy improves the sphincter preservation rate in patients with rectal cancer located within 3 cm of the anal verge. Eur J Surg Oncol. 2006;32(2):162–7. 5. National Comprehensive Cancer Network. NCCN clinical practice guidelines in oncology: rectal cancer. 2013. http://www.nccn.org/ professionals/physician_gls/pdf/rectal.pdf. Accessed 31 Dec 2014. 6. Compton CC, Fielding LP, Burgart LJ, et al. Prognostic factors in colorectal cancer: College of American Pathologists consensus statement 1999. Arch Pathol Lab Med. 2000;124(7):979–94. 7. Nelson H, Petrelli N, Carlin A, et al. Guidelines 2000 for colon and rectal cancer surgery. J Natl Cancer Inst. 2001;93(8):583–96. 8. Gunderson LL, Jessup JM, Sargent DJ, Greene FL, Stewart A. Revised tumor and node categorization for rectal cancer based on Surveillance, Epidemiology, and End Results and rectal pooled analysis outcomes. J Clin Oncol. 2010;28(2):256–63. 9. Tepper JE, O’Connell MJ, Niedzwiecki D, et al. Impact of number of nodes retrieved on outcome in patients with rectal cancer. J Clin Oncol. 2001;19:157–63.
10. Le M, Nelson R, Lee W, et al. Evaluation of lymphadenectomy in patients receiving neoadjuvant radiotherapy for rectal adenocarcinoma. Ann Surg Oncol. 2012;19:3713–8. 11. Scabini S, Montecucco F, Nencioni A, et al. The effect of preoperative chemoradiotherapy on lymph nodes harvested in TME for rectal cancer. World J Surg Oncol. 2013;11:292. 12. Govindarajan A, Go¨nen M, Weiser MR, et al. Challenging the feasibility and clinical significance of current guidelines on lymph node examination in rectal cancer in the era of neoadjuvant therapy. J Clin Oncol. 2011;29(34):4568–73. 13. Miller ED, Robb BW, Cummings OW, Johnstone PA. The effects of preoperative chemoradiotherapy on lymph node sampling in rectal cancer. Dis Colon Rectum. 2012;55(9):1002–7. 14. National Cancer Institute. Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov). SEER*Stat Database: Incidence – SEER 18 Regs Research Data, Nov 2012 Sub (1973-2010) \ Katrina/Rita Population Adjustment [— Linked To County Attributes—Total U.S., 1973-2012 Counties, National Cancer Institute, DCCPS, Surveillance Research Program, Surveillance Systems Branch, released September 2013. 15. Harrell FE Jr (2014) Regression modeling strategies with applications to linear models, logistic regression, and survival analysis. Springer, New York pp 13–15; 509–522 16. Edge SB, Byrd DR, Compton CC, Fritz AG, Greene FL, Trotti A. (eds). (2009) AJCC Cancer Staging Manual. 7th ed. Springer, New York pp 144–166 17. Wichmann MW, Mu¨ller C, Meyer G, et al. Effect of preoperative radiochemotherapy on lymph node retrieval after resection of rectal cancer. Arch Surg. 2002;137(2):206–10. 18. Baxter NN, Morris AM, Rothenberger DA, Tepper JE. Impact of preoperative radiation for rectal cancer on subsequent lymph node evaluation: a population-based analysis. Int J Radiat Oncol Biol Phys. 2005;61(2):426–31. 19. Ha YH, Jeong SY, Lim SB, et al. Influence of preoperative chemoradiotherapy on the number of lymph nodes retrieved in rectal cancer. Ann Surg. 2010;252(2):336–40. 20. Lykke J, Roikjær O, Jess P, et al. Tumour stage and preoperative chemoradiotherapy influence the lymph node yield in stages I-III rectal cancer: results from a prospective nationwide cohort study. Colorectal Dis. 2014;16(4):144–9. 21. Mekenkamp LJ, van Krieken JH, Marijnen CA, et al. Lymph node retrieval in rectal cancer is dependent on many factors: the role of the tumor, the patient, the surgeon, the radiotherapist, and the pathologist. Am J Surg Pathol. 2009;33(10):1547–53. 22. Morris EJ, Maughan NJ, Forman D, Quirke P. Identifying stage III colorectal cancer patients: the influence of the patient, surgeon, and pathologist. J Clin Oncol. 2007;25(18):2573–9. 23. Thompson RH, Carver BS, Bosl GJ, et al. Body mass index is associated with higher lymph node counts during retroperitoneal lymph node dissection. Urology. 2012;79(2):361–4. 24. van Gijn W, Marijnen CA, Nagtegaal ID, et al. Preoperative radiotherapy combined with total mesorectal excision for resectable rectal cancer: 12-year follow-up of the multicenter, randomised controlled TME trial. Lancet Oncol. 2011;12(6):575– 82. 25. Kapiteijn E, Kranenbarg EK, Steup WH, et al. Total mesorectal excision (TME) with or without preoperative radiotherapy in the treatment of primary rectal cancer: prospective randomised trial with standard operative and histopathological techniques. Eur J Surg. 1999;165(5):410–20. 26. Demetter P, Vandendael T, Sempoux C, et al. Need for objective and reproducible criteria in histopathological assessment of total mesorectal excision specimens: lessons from a national improvement project. Colorectal Dis. 2013;15(11):1351–8. 27. La Torre M, Mazzuca F, Ferri M, et al. The importance of lymph node retrieval and lymph node ratio following preoperative
M. D. Hall et al.
28.
29.
30.
31.
32.
chemoradiation of rectal cancer. Colorectal Dis. 2013;15(7):e382–8. de Campos-Lobato LF, Stocchi L, de Sousa JB, et al. Less than 12 nodes in the surgical specimen after total mesorectal excision following neoadjuvant chemoradiation: it means more than you think! Ann Surg Oncol. 2013;20(11):3398–406. Persiani R, Biondi A, Gambacorta MA, et al. Prognostic implications of the lymph node count after neoadjuvant treatment for rectal cancer. Br J Surg. 2014;101(2):133–42. Smith DD, Schwarz RR, Schwarz RE. Impact of total lymph node count on staging and survival after gastrectomy for gastric cancer: data from a large US-population database. J Clin Oncol. 2005;23(28):7114–24. Marchet A, Mocellin S, Ambrosi A, et al. The prognostic value of N-ratio in patients with gastric cancer: validation in a large, multicenter series. Eur J Surg Oncol. 2008;34(2):159–65. Wang J, Dang P, Raut CP, et al. Comparison of a lymph node ratio-based staging system with the 7th AJCC system for gastric
33.
34.
35.
36.
37.
cancer: analysis of 18,043 patients from the SEER database. Ann Surg. 2012;255(3):478–85. Wang J, Hassett JM, Dayton MT, Kulaylat MN. The prognostic superiority of log odds of positive lymph nodes in stage III colon cancer. J Gastrointest Surg. 2008;12(10):1790–6. Smith DD, Nelson RA, Schwarz RE. A comparison of five competing lymph node staging schemes in a cohort of resectable gastric cancer patients. Ann Surg Oncol. 2014;21(3):875–82. Marks JH, Valsdottir EB, Rather AA, Nweze IC, Newman DA, Chernick MR. Fewer than 12 lymph nodes can be expected in a surgical specimen after high-dose chemoradiation therapy for rectal cancer. Dis Colon Rectum. 2010;53(7):1023–9. Lee W, Nelson R, Akmal Y, et al. Racial and ethnic disparities in outcomes with radiation therapy for rectal adenocarcinoma. Int J Colorectal Dis. 2012;27(6):737–49. Kim J, Sun CL, Mailey B, et al. Race and ethnicity correlate with survival in patients with gastric adenocarcinoma. Ann Oncol. 2010;21(1):152–60.
ORIGINAL ARTICLE – COLORECTAL CANCER
Impact of Total Lymph Node Count on Staging and Survival After Neoadjuvant Chemoradiation Therapy for Rectal Cancer Matthew D. Hall, MD, MBA1, Timothy E. Schultheiss, PhD1, David D. Smith, PhD2, Marwan G. Fakih, MD3, Joseph Kim, MD4, Jeffrey Y. C. Wong, MD1, and Yi-Jen Chen, MD, PhD1 Department of Radiation Oncology, City of Hope National Medical Cancer, Duarte, CA; 2Division of Biostatistics, City of Hope National Medical Cancer, Duarte, CA; 3Department of Medical Oncology and Therapeutics Research, City of Hope National Medical Cancer, Duarte, CA; 4Department of Surgery, City of Hope National Medical Cancer, Duarte, CA 1
ABSTRACT Purpose. Current guidelines recommend that a minimum of 12 lymph nodes (LNs) be dissected to accurately stage rectal cancer patients. Neoadjuvant chemoradiation therapy (CRT) decreases the number of LNs retrieved at surgery. The purpose of this study was to assess the impact of the number of LNs dissected on overall survival (OS) for localized rectal cancer patients treated with neoadjuvant CRT. Methods. Treatment data were obtained on all patients treated for rectal cancer (2000–2013) in the National Oncology Data AllianceTM, a proprietary database of merged tumor registries. Eligible patients were treated with neoadjuvant CRT followed by surgery and had complete data on number of positive LNs, number of LNs examined, and treatment dates (n = 4565). Results. Hazard ratios for OS decreased sequentially with increasing number of LNs examined until a maximum benefit was achieved with examination of eight LNs. On multivariate analysis, age, sex, race, marital status, grade, ypT stage, ypN stage, type of surgery, margin status, presence of pathologically confirmed metastasis at surgery, and number of LNs examined were significant predictors of OS. Conclusions. Examination of eight or more LNs in rectal cancer patients treated with neoadjuvant CRT resulted in accurate staging and assignment into prognostic groups with an ensuing improvement in OS by stage. This study suggests that eight LNs is the threshold for an adequate lymph node dissection after neoadjuvant CRT.
Accurate identification of the extent of disease is essential in the diagnosis and treatment of any malignancy. Prognosis for localized rectal cancer is primarily determined by the pathologic tumor (pT) stage and the number of positive lymph nodes (LNs). Neoadjuvant chemoradiation therapy (CRT) followed by total mesorectal excision (TME) is preferred over TME and adjuvant CRT in T3–T4 and/or node-positive rectal tumors due to tumor downsizing, increased sphincter preservation rates, and reductions in local recurrence and treatment-related complications.1–4 The current National Comprehensive Cancer Network (NCCN) guidelines recommend that a minimum of 12 LNs be dissected to accurately stage and forecast prognosis in patients with rectal cancer.5–9 However, neoadjuvant CRT decreases the number of LNs retrieved at surgery, and large series indicate that fewer than 50 % of patients achieve this level of dissection.10,11 The current guidelines to dissect 12 or more LNs, based on data collected with upfront surgery, have been questioned.12,13 The optimal number of LNs dissected after neoadjuvant CRT remains poorly characterized. The purpose of this study was to assess the impact on overall survival (OS) of the number of LNs dissected in rectal cancer patients treated with neoadjuvant CRT followed by surgery using a large database and to quantify the effects of prognostic factors. METHODS AND MATERIALS Patient Selection
Ó Society of Surgical Oncology 2015 First Received: 3 February 2015 M. D. Hall, MD, MBA e-mail: [email protected]
Data on all patients diagnosed with rectal cancer from 2000 through 2013 were extracted from the National Oncology Data AllianceTM (NODA), a proprietary database of merged tumor registries (Elekta/IMPAC Medical Systems,
M. D. Hall et al.
Inc., Sunnyvale, CA, USA). The NODA captures newly diagnosed cancer cases at more than 150 hospitals in the US and is fully compliant with American College of Surgeons (ACOS) regulatory requirements. Data in the NODA registries contain exactly the same data submitted to state tumor registries and to the Surveillance, Epidemiology, and End Results (SEER) database, in regions that participate in SEER.14 The study population included all patients treated with neoadjuvant CRT followed by surgery, with complete data on the number of LNs dissected, the number of positive LNs, and the treatment start dates with chemotherapy, radiation therapy, and surgery. Patients who underwent shortcourse neoadjuvant radiation therapy, had metastatic disease at diagnosis, or had a history of prior malignancy were excluded, leaving 4565 evaluable patients. Patient characteristics included age, sex, race, marital status, clinical tumor (cT) and nodal (cN) stage, post-therapy pathologic tumor stage (ypT stage), grade, surgical margin status, presence of pathologically confirmed metastatic disease at surgery, year of diagnosis, year of treatment, number of LNs examined, number of positive LNs, type of surgery performed, radiation therapy dose, concurrent chemotherapy agent, and medical comorbidities. This study was conducted with the approval of the Institutional Review Board. The primary objective was to determine the number of examined LNs associated with the lowest mortality risk after trimodality therapy. Participating institutions in the NODA regularly update patient files with vital status for OS and cause-specific survival (CSS). Statistical Analysis Statistical analysis was performed using SPSS version 18.0 (SPSS Inc., Chicago, IL, USA) after patient-identifiers were removed from the data set. Differences between categorical variables were compared using the v2 test, and the t test was used for continuous variables. Multivariate proportional hazards analysis (MVA) with backward stepwise regression was used to identify factors associated with OS.15 Two-sided p values \0.05 were considered statistically significant. The threshold for the number of LNs examined was determined by examining the trend of the hazard ratios (HRs) on MVA and identifying naturallyoccurring breakpoints with similar survival rates. The proportional hazards assumption was tested for a non-zero slope in the generalized linear model using scaled Schoenfeld residuals. Survival curves were plotted using the Kaplan–Meier method, and compared using the logrank test.
RESULTS Patient Characteristics and the Distribution of Lymph Nodes (LNs) Examined Patient demographics and tumor characteristics are presented in Table 1. Median age was 61 years (interquartile range [IQR] 51–68). Based on the results of the threshold analysis, significant differences in clinical characteristics between patients with \8 and C8 LNs examined are also shown. Patients with \8 LNs examined were more likely to be of older age and be treated with abdominoperineal resection (APR) versus low anterior resection (LAR) than patients with C8 LNs dissected. All other comparisons showed no significant differences between the groups, or found more favorable attributes in patients with \8 LNs examined. Median follow-up was 3.5 years for surviving patients. At the close-out date, 26 % of patients had died. The median number of LNs examined was 11 (IQR 6– 16). In total, 53.1 % of patients had \12 LNs dissected. Multivariate Analysis and Threshold Analysis for Number of LNs Examined Multivariate Cox models were fit for OS, and regression diagnostics showed no violations of the proportional hazards assumptions. On MVA, age, sex, race, marital status, grade, ypT stage, post-therapy pathologic nodal stage (ypN stage), type of surgery performed, surgical margin status, presence of pathologically confirmed metastasis at the time of surgery, and number of LNs examined were significant predictors of OS (Table 2). Treatment year, cT stage, and cN stage were not significant. Exclusion of patients with pathologically confirmed metastasis at surgery (108 cases; 2.4 %) caused no significant change in the results, and this parameter was included in the model. Adjusting for the effect of other prognostic factors, number of LNs examined was a significant predictor of OS on MVA (p = 0.008). HRs for OS decreased sequentially with increasing number of LNs examined until a maximum benefit was achieved with examination of eight LNs (Table 2). Figure 1a illustrates that no further reduction in the HRs for OS was observed after eight LNs were examined. To test the validity of this finding, we separately examined the subgroup of patients with zero positive LNs (n = 3160) and only one positive LN (n = 493). MVA was repeated for each subgroup, and the results are shown in Fig. 1b and c, respectively. A similar pattern was observed, indicating that an adequate nodal dissection must contain at least eight LNs.
Lymph Node Count and Survival in Rectal Cancer TABLE 1 Patient demographics in relation to number of lymph nodes examined Characteristics
Total
Lymph nodes retrieved \8
C8
4565
1511
3054
Median
61
62
60
Interquartile range
52–69 53–70
51–68
Male
63.2
62.7
63.5
Female
36.8
37.3
36.5
No. of patients
TABLE 1 continued Characteristics
Total
p Value
Age, years \0.01
Sex
Race 0.46
\8
C8
Negative
91.4
90.7
91.7
Positive
5.9
5.4
6.2
Unknown
2.7
3.9
2.1
97.6
97.6
97.6
2.4
2.4
2.4
5-FU or capecitabine
84.9
84.8
85.0
Other/unspecified
15.1
15.2
15.0
Metastasis found at surgery No Yes
0.62
Lymph nodes retrieved
88.6
88.6
88.6
Black
6.6
6.8
6.5
\45
2.5
3.0
2.3
Asian/American Indian
3.2
3.6
3.0
45.0–50.4
67.0
65.8
67.7
Unknown
1.6
1.0
1.9
Single Married
32.0 60.2
29.0 62.8
33.4 58.9
Unknown
7.8
8.2
7.7
0.01
3.1
4.3
2.5
T2
9.4
10.8
8.7
T3
64.7
58.9
67.5
T4
4.2
4.9
3.9
Unknown
18.6
21.1
17.4
0
47.2
53.5
44.1
1
30.5
23.5
34.1
2
3.6
1.8
4.5
Unknown
18.7
21.2
17.3
1 2
7.1 66.8
6.4 69.3
7.4 65.6
3
10.9
9.4
11.6
4
0.5
0.3
0.7
Unknown
14.7
14.6
14.7
0.96
0.78
Radiation therapy dose (Gy) 0.13
50.41–54.0
12.0
12.0
12.0
[54
4.0
3.8
4.0
Unknown
14.5
15.4
14.0
Cardiovascular disease
25.0
22.4
26.3
Diabetes mellitus
7.0
6.6
7.2
0.42
Respiratory disease
2.8
2.2
3.1
0.08
Medical comorbidities
cT stage T1
0.31
Concurrent chemotherapy
White
Marital status
p Value
\0.01
\0.01
Data are expressed as percentages unless otherwise stated; two-sided p values \0.05 were considered statistically significant cT stage clinical tumor stage, cN stage clinical nodal stage, ypT stage pathologic tumor stage after neoadjuvant therapy, 5-FU 5-fluorouracil
cN stage \0.01
Grade 0.04
ypT stage ypT0
13.6
15.9
12.4
ypT1–2
30.3
32.7
29.0
ypT3
45.6
39.9
48.5
ypT4
3.3
3.0
3.5
Unknown
7.2
8.5
6.6
\0.01
Type of surgery Low anterior resection
43.9
40.3
45.7
Abdominoperineal resection 33.2
35.5
32.0
Unknown
24.2
22.3
Surgical margin
22.9
\0.01
Five-year OS for the entire cohort was 70.2 %, and the median OS was 9.7 years (95 % CI 8.9–10.5 years). For patients with \8 LNs examined, median OS was 9.4 years (95 % CI 8.6–10.2 years), and 11.1 years (95 % CI 9.5– 12.7 years) for patients with C8 LNs examined. On MVA, patients with C8 LNs examined also experienced a significant CSS advantage compared with patients with \8 LNs dissected (HR 0.87; 95 % CI 0.76–0.99, p = 0.04). Survival Outcomes Based on Pathologic Nodal Involvement Compared to patients with zero positive nodes, the HRs for OS increased with increasing number of positive nodes when divided into five nodal tiers as per the current American Joint Committee on Cancer (AJCC) guidelines (Table 2).16 The Kaplan–Meier curves showed a clear and statistically significant difference in OS in patients with 0, 1, 2–3, 4–6, and C 7 positive LNs (p \ 0.0001), as shown in Fig. 2a. This pattern was also observed for CSS (p \ 0.0001), as shown in Fig. 2b. In the subset of patients
M. D. Hall et al.
(a)
TABLE 2 Hazard ratios for overall survival p Value
HR
95 % CI
All Patients 1.3 1.2 1.1
Number of lymph nodes examined Hazard Ratio
1.0
0–1
0.008
: 1.00
2–3
0.832
0.97
0.73–1.28
4–5
0.401
0.89
0.67–1.17
6–7
0.373
0.88
0.68–1.15
8
0.029
0.69
0.50–0.96
0.4
9–10
0.020
0.71
0.53–0.95
0.3
11–12
0.023
0.71
0.53–0.96
13–14
0.016
0.69
0.52–0.94
C15
0.005
0.69
0.54–0.89
\0.001
1.03
1.02–1.04
0.7 0.6
\0.001
–
–
1.1
0.73
0.64–0.83
1.0
9 ≥1
-1 2 13 -1 4 15 -1 6 17 -1 8
11
8 910
6
7
4
0.8
–
Black
\0.001
1.59
1.28–1.97
American Indian
0.051
1.56
1.00–2.44
0.5
Asian
0.994
1.00
0.62–1.62
0.4
0.7 0.6
0.3
Marital status
1.15
0.89–1.47
ypT3
\0.001
1.86
1.47–2.36
0.2
ypT4
\0.001
2.71
1.93–3.81
0.0
\0.001
–
–
9-
Number of Lymph Nodes Examined
(c)
One Positive Lymph Nodes 1.4
1.0 0.8 0.6
\0.001
1.44
1.20–1.73
\0.001
1.82
1.52–2.18
ypN2a
\0.001
2.25
1.80–2.81
ypN2b
\0.001
3.39
2.70–4.25
Type of surgery – \0.001
–
–
1.28
1.12–1.47
8 ≥1 9
5
-1 16
2
-1
-1 10
ypN1a ypN1b
13
9 8-
7
6
0.4
2
Hazard Ratio
1.2
1-
ypN stage
9
0.293
8
–
\0.001
≥1
–
ypT stage
6
1.02–3.93
-1
2.00
17
0.044
4
4
-1
1.17–2.09
-1
0.92–1.54
1.57
15
1.19
0.002
13
0.196
3
2
2
-1
–
11
–
8
0.001
10
1
7
0.73–0.94
6
0.83
5
0.005
4
Married Grade
3
–
2
–
0
–
1
0.2
Unmarried
Abdominoperineal resection
5
0.9
–
Low anterior resection
Zero Positive Lymph Nodes
1.2
–
\0.001
ypN0
3
(b)
White
ypT1–T2
2
Number of Lymph Nodes Examined
1.3
Race
ypT0
1
0
0.2
Hazard Ratio
Female
0.8
0.5
Sex Male
0.9
35
Age (Continuous)
–
Number of Lymph Nodes Examined
FIG. 1 Hazard ratios for overall survival by number of lymph nodes examined on multivariate analysis. Adjusting for the effect of other factors, hazard ratios for overall survival declined with an increasing number of lymph nodes dissected in (a) all patients, (b) patients with negative lymph nodes, and (c) patients with only one positive lymph node. No further reduction in the hazard ratio for overall survival was observed after eight lymph nodes were examined. The error bars indicate the standard error of the hazard ratio
Surgical margin Negative Positive
– \0.001
–
–
1.62
1.33–1.98
–
–
3.03
2.37–3.89
Metastasis found at surgery No Yes
– \0.001
HR hazard ratio, CI confidence interval, ypT stage pathologic tumor stage after neoadjuvant therapy, ypN stage pathologic nodal stage after neoadjuvant therapy
with C8 LNs dissected, MVA again demonstrated that the number of positive LNs was significant with increasing HRs for OS in patients with 0, 1, 2–3, 4–6, and C7 positive nodes (p \ 0.001). As an exploratory analysis, MVA was performed with number of positive LNs included as a categorical variable to examine the discrimination utility of the current AJCC
Lymph Node Count and Survival in Rectal Cancer All Patients
1.0
Overall Survival
0.8
AJCC N stage N0 N1a N1b N2a N2b
0.6
0.4
4.0
Hazard Ratio
(a)
3.0
2.0
0.2
1.0
0.0 0
Number of Positive Lymph Nodes
Number at Risk 0 LNs Positive
3,160
2,719
2,234
1,774
1,372
1,019
746
1 LN Positive
493
426
342
273
211
158
108
2-3 LNs Positive
457
390
309
225
170
116
83
4-6 LNs Positive
252
207
162
114
67
44
35
≥7 LNs Positive
203
168
116
77
53
33
17
Cause-Specific Survival
(b)
≥8
6
7
5
6
4
5
3
Time (Years)
4
2
3
1
2
0
1
0.0
FIG. 3 Hazard ratios for overall survival by number of positive lymph nodes on multivariate analysis. Groups with similar hazard ratios were combined, yielding five nodal groupings (0, 1, 2–4, 5–7, and C8 positive lymph nodes) that best forecast prognosis in this cohort. The error bars indicate the standard error of the hazard ratio
1.0
0.8
AJCC N stage N0 N1a N1b N2a N2b
0.6
0.4
no additional prognostic information than ypN stage if C8 nodes were examined; ypN stage remained the more useful prognostic variable if \8 nodes were retrieved. DISCUSSION
0.2
0.0 0
1
2
3
4
5
6
Time (Years) Number at Risk 0 LNs Positive
3,110
2,631
2,085
1,600
1,217
891
650
1 LN Positive
488
407
304
227
173
125
93
2-3 LNs Positive
451
365
260
187
137
90
66
4-6 LNs Positive
252
195
137
85
53
33
28
≥7 LNs Positive
199
148
85
49
31
19
10
FIG. 2 (a) Overall survival and (b) cause-specific survival by number of positive lymph nodes. The Kaplan–Meier curves showed a clear and statistically significant association between number of positive lymph nodes and improved overall survival (p \ 0.0001) and cause-specific survival (p \ 0.0001). AJCC American Joint Committee on Cancer, LNs lymph nodes
nodal groupings. Adjusting for the effect of other significant variates, Fig. 3 shows that the HRs for OS increased monotonically in patients with 0, 1, 2–4, 5–7, and C8 positive nodes. When adjoining groups with similar survival rates were merged, Kaplan–Meier analysis again demonstrated a clear and statistically significant difference in OS (p \ 0.0001). For ypN staging, this nodal grouping showed higher discrimination utility as measured by the v2 statistic compared with the AJCC system in this dataset. LN ratio, with patients divided into three tiers (0, 0–30, and [30 %), was significant on MVA when ypN stage was not included in the model. When ypN stage was included, LN ratio was not significant (p * 0.3). LN ratio provided
An adequate LN dissection is critical to delineate the true stage of disease. Published guidelines recommend that a minimum of 12 LNs be dissected to accurately forecast prognosis in patients with colorectal cancer, but did not consider the effects of neoadjuvant therapy.5–7 In rectal cancer, neoadjuvant CRT is employed in appropriately selected patients to improve locoregional control and functional outcomes,1–4 but also significantly reduces the number of LNs retrieved compared to those treated with surgery alone.10,11,17–20 The optimal number of LNs needed to accurately stage and forecast prognosis in patients treated with neoadjuvant CRT has not been previously characterized. In a large patient cohort, our study demonstrated that the number of LNs examined after neoadjuvant CRT was a significant prognostic factor and that OS improved with increasing number of nodes examined until a maximum benefit was achieved with examination of eight LNs. This pattern was also observed on subset analysis in patients with zero positive LNs and only one positive LN. As a result, we recommend that eight or more LNs be dissected to improve the accuracy of nodal staging. Although a target for the minimum number of examined LNs after neoadjuvant CRT was identified, it is important to note that many patients with rectal cancer fail to achieve this level of nodal dissection. In our cohort, 33.1 % of patients had \8 LNs examined. Only 46.9 % met the current guidelines of C12 LNs examined, which is
M. D. Hall et al.
consistent with published multi-institutional and population-based studies.10,18,19,21 In addition to neoadjuvant therapy, patient and disease-specific factors, including age, sex, body mass index (BMI), tumor location, and pT stage, have been reported to impact the number of LNs collected after pelvic surgery.18–23 TME versus proctectomy alone, as well as the experience of the surgeons and pathologists involved, have also been shown to influence the LN yield.22 Even so, TME remains a technically challenging surgery and up to 25 % of cases have an incomplete resection of the mesorectum, even in the hands of experts.24–26 The reason for the pervasiveness of inadequate nodal dissections in rectal cancer is multifactorial and this trend is unlikely to be reversed in the near-term. However, knowledge regarding the number of LNs needed to adequately stage patients treated with neoadjuvant CRT is an important first step. In our study, a five-tier nodal division was found to best forecast prognosis based on number of positive LNs identified after neoadjuvant CRT, consistent with the current AJCC guidelines. In our analysis, the HRs for OS increased monotonically in patients with 0, 1, 2–4, 5–7, and C8 positive nodes. These groupings were determined by examining the HRs for number of positive LNs on MVA (Fig. 3) and merging adjoining groups with similar survival rates. The resulting five-tier system marginally diverges from the AJCC 7th edition nodal staging (0, 1, 2–3, 4–6, and C7 positive LNs). In our cohort, the five-tier nodal grouping we identified had superior discrimination utility and better predicted OS, based on the number of positive LNs, than the current AJCC system. Five-year OS for our study was 77 % for patients with negative LNs, 67 and 58 % for patients with 1 and 2–4 positive LNs, respectively, and 38 and 30 % for patients with 5–7 and C8 positive LNs, respectively. We recommend that this new grouping be considered for the next version of the AJCC pathologic nodal (pN) strata for rectal cancer. This study suggests that the number of LNs examined is an important prognostic parameter in rectal cancer patients treated with neoadjuvant CRT. Prior studies have not demonstrated an association between the number of LNs examined and OS after neoadjuvant CRT.12,27–29 Although single institution series from tertiary care centers may possess more uniform surgical and pathologic technique, sample size may limit their ability to detect smaller effect sizes due to reduced statistical power. Larger sample size is an advantage of pooled registries and population-based databases. Several limitations in our study must be acknowledged. Although this study is the largest to address the impact of LN involvement in patients treated with neoadjuvant CRT followed by surgery for rectal cancer, the data do not enable the opportunity to examine several important factors,
including surgical technique, use of and adherence to adjuvant chemotherapy, and postoperative complications. In addition, the observed detriment in OS due to APR may be related to correlated variables that could not be analyzed, including proximal versus distal tumor location. The outcome measures were limited to OS and CSS, and data regarding local and distant failure cannot be extracted. Patient selection can influence the results in populationbased observational series and must be considered when applying our results to clinical practice. However, the completeness of the available data helps to counter the possibility that the observed LN threshold is due to selection bias. First, the analysis included many known variables that can influence OS in rectal cancer patients, including surgical margin status and even the presence of biopsy-proven metastatic disease at surgery, which are not available in the SEER database. Second, patient demographics and reported medical comorbidities in the subgroup of patients with \8 and C8 LNs dissected were similar, or even favored the group with fewer LNs dissected (Table 1). This suggests that the observed benefit afforded by an ‘adequate’ LN dissection was not the product of surgery provided to inherently healthier individuals. Finally, we acknowledge that observational studies cannot replace randomized data as the standard for outcomes research. However, population-based analyses can provide measures of the effectiveness of therapies in the general population and address important clinical questions that have not been adequately evaluated in randomized trials. Stage migration, due to inappropriate understaging of disease, is an important mechanism to explain the difference in OS observed. Given that the number of positive LNs identified is fundamentally related to the number of LNs examined, a fraction of patients classified as N0 will be understaged.30 This effect can be mitigated by examining a larger number of LNs, specifically C8 in this analysis. In addition, alternative conventions, including LN ratio and the logarithmic odds approach, have been proposed to improve on the existing AJCC staging criteria.31–34 Despite the large body of literature on this topic, as well as our findings, several important questions remain. In this study, cT and cN stage were not significant on MVA after the effect of other variates were included in the model. Earlier work has suggested that the response to neoadjuvant therapy for pT stage and LN yield may not be highly correlated.20,35 Tumor regression is an important metric, and the impact of neoadjuvant CRT on subsequent pT stage and nodal positivity is an important outcome to be further explored. The identification of patients that may be candidates for transanal excision based on objective tumor response criteria, especially in less than optimal surgical candidates, is also of great interest. Finally, in our study,
Lymph Node Count and Survival in Rectal Cancer
the difference in OS for Black and American Indian patients relative to White patients has previously been reported in rectal and other gastrointestinal malignancies. Socioeconomic factors, access to care, and surgical technique have been posited as potential explanations for observed racial disparities but have not been supported by the data.36,37 Racial disparities may also be a function of differing tumor biology, and the molecular and genetic differences in rectal cancer also warrant further investigation. CONCLUSIONS In a large, population-based cohort, our study demonstrates that examination of eight or more LNs in rectal cancer patients treated with neoadjuvant CRT effected accurate staging and allocation into prognostic groups with a resulting improvement in OS. This suggests that eight LNs is the threshold for an adequate nodal dissection after neoadjuvant CRT. A five-tier nodal grouping was found to best forecast prognosis based on the number of positive LNs identified after neoadjuvant CRT. This grouping (0, 1, 2–4, 5–7, and C8 positive LNs) should be considered in the next version of the AJCC pN staging for rectal cancer. DISCLOSURE
None.
REFERENCES 1. Sauer R, Becker H, Hohenberger W, et al. Preoperative versus postoperative chemoradiotherapy for rectal cancer. N Engl J Med. 2004;351(17):1731–40. 2. Peeters KC, Marijnen CA, Nagtegaal ID, et al. The TME trial after a median follow-up of 6 years: increased local control but no survival benefit in irradiated patients with resectable rectal carcinoma. Ann Surg. 2007;246(5):693–701. 3. Ortholan C, Francois E, Thomas O, Benchimol D, Baulieux J, Bosset JF, et al. Role of radiotherapy with surgery for T3 and resectable T4 rectal cancer: evidence from randomized trials. Dis Colon Rectum. 2006;49(3):302–10. 4. Kim DW, Lim SB, Kim DY, et al. Pre-operative chemo-radiotherapy improves the sphincter preservation rate in patients with rectal cancer located within 3 cm of the anal verge. Eur J Surg Oncol. 2006;32(2):162–7. 5. National Comprehensive Cancer Network. NCCN clinical practice guidelines in oncology: rectal cancer. 2013. http://www.nccn.org/ professionals/physician_gls/pdf/rectal.pdf. Accessed 31 Dec 2014. 6. Compton CC, Fielding LP, Burgart LJ, et al. Prognostic factors in colorectal cancer: College of American Pathologists consensus statement 1999. Arch Pathol Lab Med. 2000;124(7):979–94. 7. Nelson H, Petrelli N, Carlin A, et al. Guidelines 2000 for colon and rectal cancer surgery. J Natl Cancer Inst. 2001;93(8):583–96. 8. Gunderson LL, Jessup JM, Sargent DJ, Greene FL, Stewart A. Revised tumor and node categorization for rectal cancer based on Surveillance, Epidemiology, and End Results and rectal pooled analysis outcomes. J Clin Oncol. 2010;28(2):256–63. 9. Tepper JE, O’Connell MJ, Niedzwiecki D, et al. Impact of number of nodes retrieved on outcome in patients with rectal cancer. J Clin Oncol. 2001;19:157–63.
10. Le M, Nelson R, Lee W, et al. Evaluation of lymphadenectomy in patients receiving neoadjuvant radiotherapy for rectal adenocarcinoma. Ann Surg Oncol. 2012;19:3713–8. 11. Scabini S, Montecucco F, Nencioni A, et al. The effect of preoperative chemoradiotherapy on lymph nodes harvested in TME for rectal cancer. World J Surg Oncol. 2013;11:292. 12. Govindarajan A, Go¨nen M, Weiser MR, et al. Challenging the feasibility and clinical significance of current guidelines on lymph node examination in rectal cancer in the era of neoadjuvant therapy. J Clin Oncol. 2011;29(34):4568–73. 13. Miller ED, Robb BW, Cummings OW, Johnstone PA. The effects of preoperative chemoradiotherapy on lymph node sampling in rectal cancer. Dis Colon Rectum. 2012;55(9):1002–7. 14. National Cancer Institute. Surveillance, Epidemiology, and End Results (SEER) Program (www.seer.cancer.gov). SEER*Stat Database: Incidence – SEER 18 Regs Research Data, Nov 2012 Sub (1973-2010) \ Katrina/Rita Population Adjustment [— Linked To County Attributes—Total U.S., 1973-2012 Counties, National Cancer Institute, DCCPS, Surveillance Research Program, Surveillance Systems Branch, released September 2013. 15. Harrell FE Jr (2014) Regression modeling strategies with applications to linear models, logistic regression, and survival analysis. Springer, New York pp 13–15; 509–522 16. Edge SB, Byrd DR, Compton CC, Fritz AG, Greene FL, Trotti A. (eds). (2009) AJCC Cancer Staging Manual. 7th ed. Springer, New York pp 144–166 17. Wichmann MW, Mu¨ller C, Meyer G, et al. Effect of preoperative radiochemotherapy on lymph node retrieval after resection of rectal cancer. Arch Surg. 2002;137(2):206–10. 18. Baxter NN, Morris AM, Rothenberger DA, Tepper JE. Impact of preoperative radiation for rectal cancer on subsequent lymph node evaluation: a population-based analysis. Int J Radiat Oncol Biol Phys. 2005;61(2):426–31. 19. Ha YH, Jeong SY, Lim SB, et al. Influence of preoperative chemoradiotherapy on the number of lymph nodes retrieved in rectal cancer. Ann Surg. 2010;252(2):336–40. 20. Lykke J, Roikjær O, Jess P, et al. Tumour stage and preoperative chemoradiotherapy influence the lymph node yield in stages I-III rectal cancer: results from a prospective nationwide cohort study. Colorectal Dis. 2014;16(4):144–9. 21. Mekenkamp LJ, van Krieken JH, Marijnen CA, et al. Lymph node retrieval in rectal cancer is dependent on many factors: the role of the tumor, the patient, the surgeon, the radiotherapist, and the pathologist. Am J Surg Pathol. 2009;33(10):1547–53. 22. Morris EJ, Maughan NJ, Forman D, Quirke P. Identifying stage III colorectal cancer patients: the influence of the patient, surgeon, and pathologist. J Clin Oncol. 2007;25(18):2573–9. 23. Thompson RH, Carver BS, Bosl GJ, et al. Body mass index is associated with higher lymph node counts during retroperitoneal lymph node dissection. Urology. 2012;79(2):361–4. 24. van Gijn W, Marijnen CA, Nagtegaal ID, et al. Preoperative radiotherapy combined with total mesorectal excision for resectable rectal cancer: 12-year follow-up of the multicenter, randomised controlled TME trial. Lancet Oncol. 2011;12(6):575– 82. 25. Kapiteijn E, Kranenbarg EK, Steup WH, et al. Total mesorectal excision (TME) with or without preoperative radiotherapy in the treatment of primary rectal cancer: prospective randomised trial with standard operative and histopathological techniques. Eur J Surg. 1999;165(5):410–20. 26. Demetter P, Vandendael T, Sempoux C, et al. Need for objective and reproducible criteria in histopathological assessment of total mesorectal excision specimens: lessons from a national improvement project. Colorectal Dis. 2013;15(11):1351–8. 27. La Torre M, Mazzuca F, Ferri M, et al. The importance of lymph node retrieval and lymph node ratio following preoperative
M. D. Hall et al.
28.
29.
30.
31.
32.
chemoradiation of rectal cancer. Colorectal Dis. 2013;15(7):e382–8. de Campos-Lobato LF, Stocchi L, de Sousa JB, et al. Less than 12 nodes in the surgical specimen after total mesorectal excision following neoadjuvant chemoradiation: it means more than you think! Ann Surg Oncol. 2013;20(11):3398–406. Persiani R, Biondi A, Gambacorta MA, et al. Prognostic implications of the lymph node count after neoadjuvant treatment for rectal cancer. Br J Surg. 2014;101(2):133–42. Smith DD, Schwarz RR, Schwarz RE. Impact of total lymph node count on staging and survival after gastrectomy for gastric cancer: data from a large US-population database. J Clin Oncol. 2005;23(28):7114–24. Marchet A, Mocellin S, Ambrosi A, et al. The prognostic value of N-ratio in patients with gastric cancer: validation in a large, multicenter series. Eur J Surg Oncol. 2008;34(2):159–65. Wang J, Dang P, Raut CP, et al. Comparison of a lymph node ratio-based staging system with the 7th AJCC system for gastric
33.
34.
35.
36.
37.
cancer: analysis of 18,043 patients from the SEER database. Ann Surg. 2012;255(3):478–85. Wang J, Hassett JM, Dayton MT, Kulaylat MN. The prognostic superiority of log odds of positive lymph nodes in stage III colon cancer. J Gastrointest Surg. 2008;12(10):1790–6. Smith DD, Nelson RA, Schwarz RE. A comparison of five competing lymph node staging schemes in a cohort of resectable gastric cancer patients. Ann Surg Oncol. 2014;21(3):875–82. Marks JH, Valsdottir EB, Rather AA, Nweze IC, Newman DA, Chernick MR. Fewer than 12 lymph nodes can be expected in a surgical specimen after high-dose chemoradiation therapy for rectal cancer. Dis Colon Rectum. 2010;53(7):1023–9. Lee W, Nelson R, Akmal Y, et al. Racial and ethnic disparities in outcomes with radiation therapy for rectal adenocarcinoma. Int J Colorectal Dis. 2012;27(6):737–49. Kim J, Sun CL, Mailey B, et al. Race and ethnicity correlate with survival in patients with gastric adenocarcinoma. Ann Oncol. 2010;21(1):152–60.
