Magnetic resonance imaging directed surgical decision making for lateral pelvic lymph node dissection in rectal cancer after total neoadjuvant therapy (TNT)

Abstract

OBJECTIVE:

Lateral pelvic lymph node (LPLN) metastases are an important cause of preventable local failure in rectal cancer. The aim of this study was to evaluate clinical and oncological outcomes following MRI-directed surgical selection for LPLN dissection (LPLND) after TNT.

METHODS:

A retrospective consecutive cohort analysis was performed of rectal cancer patients with enlarged LPLN on pre-treatment MRI. Patients were categorized as LPLND or non-LPLND. The main outcomes were lateral local recurrence rate, perioperative and oncological outcomes and factors associated with decision-making for LPLND.

RESULTS:

A total of 158 patients with enlarged pre-treatment LPLN and treated with TNT were identified. Median follow-up was 20 months (IQR 10–32). After multidisciplinary review, 88 patients (56.0%) underwent LPLND. Mean age was 53 (SD+/− 12) years, and 54 (34.2%) were female. Total operative time (509 vs 429 minutes; p=0.003) was greater in the LPLND group, but median blood loss (p=0.70) or rates of major morbidity (19.3% vs. 17.0%) did not differ. LPLNs were pathologically positive in 34.1%. The 3-year lateral local recurrence rates (3.4% vs. 4.6%; p=0.85) did not differ between groups. Patients with LPLNs demonstrating pre-treatment heterogeneity and irregular margin (OR 3.82; 95% CIs: 1.65–8.82) or with short-axis ≥5 mm post-TNT (OR 2.69; 95% CIs: 1.19–6.08) were more likely to undergo LPLND.

CONCLUSION:

For rectal cancer patients with evidence of LPLN metastasis, the appropriate selection of patients for LPLND can be facilitated by a multidisciplinary MRI directed approach with no significant difference in perioperative or oncologic outcomes.

MINI-ABSTRACT:

For rectal cancer patients with evidence of LPLN metastasis, the appropriate selection of patients for therapeutic LPLND can be facilitated by a multidisciplinary MRI directed approach without adverse impact on perioperative or oncologic outcomes. Selection of patients for LPLND was associated with LPLNs demonstrating heterogeneity and irregular margin, and/or with a short axis LPLN ≥ 5 mm post-TNT.

INTRODUCTION:

Since the introduction of total mesorectal excision (TME) surgery and neoadjuvant radiotherapy, local recurrence rates for rectal cancer have significantly improved. Lateral local recurrence is now the most common type of local recurrence after rectal cancer surgery, which is due to lateral pelvic lymph node (LPLN) metastases(1). The management of LPLNs in rectal cancer has historically been associated with a significant divide between Western and Eastern countries. The Japanese Society for Cancer of the Colon and Rectum Guidelines recommend total mesorectal excision (TME) with lateral pelvic lymph node dissection (LPLND) as the standard procedure for locally advanced low rectal cancer(2). In contrast, Western countries consider neoadjuvant chemoradiotherapy (nCRT), and now total neoadjuvant therapy (TNT), in combination with TME as the standard treatment for locally advanced rectal cancer(3–5).

The differing paradigms have developed from competing priorities. In the West, the circumferential resection margin (CRM) has been the main focus because it is regarded as a core quality indicator of the surgical resection(6, 7) and is a strong predictor of local recurrence, distal recurrence and survival(8). This approach was supplemented by radiation and chemotherapy to treat the LPLN disease. In the East, evolution of rectal cancer surgery has emphasized complete surgical lymph node clearance, with LPLNs considered as locoregional, thus, surgical treatment has included routine dissection of this compartment for low rectal cancers (below the peritoneal reflection), without the use of radiotherapy. However, neither approach has adequately addressed the important issue of lateral compartment recurrence in patients with clinically evident LPLN metastasis(9–12).

In the West, slow adoption of LPLND was associated with particular concerns regarding the technical difficulties, especially in the more obese Western population and following nCRT(13). Furthermore, the potential for increased morbidity with LPLND, including urinary and sexual dysfunction, with LPLND and the predominantly retrospective data that did not demonstrate a clearly significant overall cancer-specific advantage, resulted in a lack of interest for LPLND in the West(3). However, the limitation of these data is that they did not distinguish between prophylactic LPLND or nCRT in the absence of clinical evidence of disease versus therapeutic LPLND in the presence of disease. Approximately 20% of locally advanced rectal cancers will present with LPLN metastasis(14) and recent data shows that these patients have a high risk of local recurrence irrespective of whether TME plus nCRT without LPLND or TME and LPLND without nCRT is performed(14–16). Therefore, the international community of rectal cancer surgeons is increasingly advocating for patients with clinically evident LPLN metastasis to be treated with a combination of both nCRT and LPLND(10, 17–19).

Although the role of therapeutic LPLND is now becoming established, there remains a need to better select who should undergo LPLND based on clinical evaluation and there is no clear consensus that exists at present. Technical improvements in magnetic resonance imaging (MRI) has greatly improved the diagnostic and staging accuracy in rectal cancer patients, facilitating identification of high-risk disease(20, 21), particularly LPLNs(22). Current NCCN guidelines recommend total neoadjuvant therapy (TNT) in the management of locally advanced rectal cancer(23). On pre-treatment MRI, the LPLNs are defined as enlarged nodes in one of the lateral nodal basins with or without malignant features, such as heterogeneity or an irregular border(24).

To date, no study has compared TME and therapeutic LPLND versus TME alone following TNT in Western rectal cancer patients with evidence of enlarged LPLN on pre-treatment MRI and it remains unclear whether LPLND after TNT leads to lower lateral local recurrence (LLR) in this population. The aim of this study was to evaluate clinical and oncological outcomes following image directed surgical decision making for LPLND in patients at a single comprehensive cancer center and to determine which patients with locally advanced rectal cancer and LPLNs treated with TNT benefit from therapeutic LPLND.

METHODS:

Patient Identification:

A retrospective cohort study of consecutive locally advanced rectal cancer patients with enlarged lateral pelvic lymph nodes (LPLN) on pre-TNT MRI was performed from a prospectively maintained database and direct medical review. This was supplemented by natural language processing review of health data obtained from structured and unstructured electronic health record elements, clinical note text and diagnostic imaging data integrated using the Palantir Foundry (Syntropy), an interoperable and extensible cloud-based platform within the MD Anderson’s Context Engine data management system(25). All patient-level data were aggregated and analyzed using these platforms to confirm complete identification of the patients. Prospectively collected data included pathology and perioperative surgical complications information. Furthermore, this data was supplemented by direct record review to assess the clinical, pathological and oncological outcomes. Patients were classified as having had LPLND or no dissection at the time of surgery. Patients who required en-bloc resection of internal iliac vessels or branches, were excluded from the analysis, as this was considered a resection of the lateral pelvic sidewall. All patients were evaluated and operated upon at a single tertiary referral center with specialist experience in the surgical management of locally advanced and recurrent rectal cancers, including LPLND, at The University of Texas MD Anderson Cancer Center (Houston, USA). This study was approved by The University of Texas MD Anderson Cancer Center’s Institutional Review Board.

Multidisciplinary Evaluation:

All cases were reviewed at a dedicated rectal cancer multidisciplinary conference, identifying those with enlarged LPLNs with biopsy-proven primary rectal adenocarcinoma, prior to initiating treatment. Preoperative staging consisted of computed tomography (CT) of the chest, abdomen and pelvis and dedicated high-resolution rectal protocol pelvic magnetic resonance imaging (MRI) without intravenous contrast and with diffusion weighted images. MRI scans were reported using a standardized reporting form(26) and TNT was recommended for patients with locally advanced tumors (involved or threatened circumferential resection margin ≤ 2mm, ≥ T3c tumors, EMVI positive, or LPLN positive(27). In some cases, neoadjuvant short-course radiotherapy was performed to minimize the duration of time without systemically active therapy, particularly for patients presenting with distant metastases or those thought to be at high-risk of systemic failure(28, 29). The radiation fields included the LPLN basins. Patients were re-evaluated with CT and MRI prior to definitive surgical resection.

MRI Assessment:

Each MRI study was re-reviewed by two participating specialty radiology co-investigators and data collected using a structured reporting proforma. The standard TNM staging, circumferential resection margin assessment, height of the tumor from the anal verge and presence of extramural vascular invasion were all recorded on the pre-treatment and post-TNT scans. In addition, assessment of the LPLN status was based on the largest LPLN identified on pre-treatment MRI. The short-axis size, site (internal iliac, obturator or common iliac), malignant signal and margin features (border) of the LPLN were also recorded. Changes in the size and presence of malignant features of the same LPLN following TNT were then evaluated.

Surgical Treatment:

Surgery was performed following completion of TNT, either at a median of 8 weeks following completion of radiation therapy or a minimum of 4 weeks following completion of consolidation chemotherapy. The patients all underwent robotic or open TME surgery, depending on the surgeon, patient or tumor factors. En-bloc multi-visceral resection was performed if there was evidence of a pre-operatively threatened or involved circumferential resection margin on imaging, as per the Beyond TME Consensus guidelines(30). All procedures included high vascular pedicle dissection with proximal ligation along with stepwise TME dissection. All patients received pre-operative bowel preparation and antibiotics prior to surgery. Patients undergoing sphincter preservation surgery had a loop ileostomy formed. The surgical technique for LPLND for rectal cancer at our institution has previously been described(13, 31, 32). Briefly, the dissection is performed along the embryological planes of the lateral pelvic compartment, identifying the key anatomical landmarks with nerve preservation. The nodal tissue within the following boundary structures are removed en-bloc: proximally (external/internal iliac bifurcation), distally (obturator foramen), laterally (external iliac vessels, psoas and obturator internus), medially (uretero-hypogastric fascia), anteriorly (peri-vesicular fat) and posteriorly (sacral nerve roots and piriformis muscle). Postoperatively, patients were managed according to the institution’s principles of enhanced recovery program for colon and rectal surgery(33).

After retrieval, specimens were taken fresh to the pathology laboratory by the operating surgeon to orient, assess and mark. In conjunction with the lead pathologist, the relevant macroscopic margins including vascular pedicle, circumferential resection margin (CRM) and distal resection margin (DRM) and the LPLNs were immediately reviewed as fresh specimens to avoid ambiguity during the final histopathologic evaluation. Pathological staging was based upon the American Joint Committee on Cancer staging guidelines, 8^th edition(34). Beyond TME surgery was defined as a surgical resection beyond conventional TME planes to achieve a negative resection margin(30). CRM and DRM involvement were defined as a resection margin of 1mm or less. Completeness of TME was also recorded, as complete, near complete and incomplete(35).

Clinical Endpoints:

The primary endpoint was lateral local recurrence (LLR) rates between the LPLND and the non-LPLND patients undergoing surgery for rectal cancer, who had evidence of enlarged LPLNs on pre-treatment MRI. Secondary endpoints included factors associated with performance of LPLND and short-term peri-operative and oncologic outcomes. Patient demographics (age, sex, BMI, ASA), comorbidities, neoadjuvant & adjuvant therapies, operative details, histopathological staging and perioperative outcome details were recorded.

Complications occurring within 30-days postoperatively were graded according to the Clavien-Dindo classification system for surgical complications(36). A major complication was defined as Grade III or IV of the Clavien-Dindo classification. Inpatient mortality was defined as a death occurring within 30-days of the index procedure. Follow-up was calculated from the date of curative surgery to the last clinic appointment.

Lateral local recurrence was defined as tumor recurrence in one of the LPLN basins and identified by cross-sectional imaging. Local recurrence was defined as tumor regrowth in the pelvis at the site of the anastomosis, previously resected mesorectal tissue planes or in one of the LPLN basins. Distant metastases were defined as tumor recurrence in distant organs or extra-regional lymph nodes. Disease free survival was defined as any disease recurrence or death from any cause.

Statistical Analysis:

Data were analyzed using STATA SE version 17·0 (StataCorp LP. College Station, Texas, USA). Continuous variable parametricity was tested using the Kolmogorov-Smirnov test and parametric data were presented as mean with standard deviation (SD) and non-parametric data were expressed as median with interquartile range (IQR). Cohort demographic and clinical characteristics were compared using Chi-squared (χ2) test or Fishers exact test for categorical variables, Mann-Whitney U test for non-parametric continuous variables and t-test for parametric continuous variables. Analysis for LLR, LR, distant recurrence and disease-free survival (DFS) were calculated using the Kaplan-Meier method. Multivariable Cox proportional hazards models were used to assess the impact of the prognostic factors of interest on survival outcomes. Univariable and multivariable analysis using logistic regression were utilized to determine possible prognostic variables (patient and tumor related factors) that determined LPLND or not. Statistical significance was considered with p values of < 0.05.

RESULTS:

Patient and treatment characteristics:

A total of 158 locally advanced rectal cancer patients with enlarged LPLN on pre-TNT MRI were identified at a single NCI designated cancer institution between January 2016 and May 2021. The median follow-up was 20 months (IQR 10–32). All patients underwent TNT, compromised of pelvic (chemo)radiotherapy with either induction or consolidation chemotherapy. The mean age was 53 (SD +/−12.0) years and 54 (34.2%) were female. Eighty-eight (56.0%) patients underwent TME and LPLND and the remaining 70 (44.0%) patients underwent TME surgery without LPLND (non-LPLND). Patient demographics and clinical characteristics are outlined in Table 1. Notably, there was no differences in the median BMI (28.4 kg/m² vs. 29.5 kg/m²; p=0.23), median tumor distance from the anal verge (6 cm vs. 6cm; p=0.63), rate of threatened mesorectal fascia margin on pre-operative imaging (55.7% vs. 55.7%; p=1.00), or rate of extramural vascular invasion (47.7% vs. 47.1%; p=0.94) in the LPLND versus the non-LPLND groups respectively.

Table 1:

Baseline patient characteristics for complete cohort of rectal cancer patients with pre-treatment abnormal LPLNs either undergoing TNT followed by LPLND or non-LPLND.

Characteristic	LPLN Non-Dissection (n = 70)	LPLN Dissection (n = 88)	p-value
Age (years) (mean, SD)	53 (+/− 10)	53 (+/− 13)	0.32
Sex (%)			0.17
Male	42 (60.0)	62 (70.5)
Female	28 (40.0)	26 (29.5)
Race (%)			0.15
Caucasian	47 (67.1)	62 (70.5)
African-American	6 (8.6)	1 (1.1)
Hispanic	13 (18.6)	17 (19.3)
Asian-American/Pacific-Islander	4 (5.7)	8 (9.1)
T categorya (%)			0.85
cT2	5 (7.1)	6 (6.8)
cT3	42 (60.0)	52 (59.1)
cT4	23 (32.9)	29 (33.0)
N categoryb (%)			0.32
cN0	3 (4.3)	1 (1.1)
cN+	67 (95.7)	86 (97.7)
M category (%)			0.28
cM1	15 (21.4)	26 (29.5)
Tumor Distance from Anal Verge (cm) (median, IQR)	6.3 (4.1–8)	6.0 (3.8–8)	0.63
Craniocaudal Length of Tumor (cm) (median, IQR)	4.9 (3.5–6.7)	5.0 (3.7–6.6)	0.71
Clinical MRF Threatenedc (%)	39 (55.7)	49 (55.7)	1.00
Extramural Vascular Invasion (%)	33 (47.1)	42 (47.7)	0.94
Baseline CEA (median, IQR)	5.5 (2.4–9.7)	4.1 (2.1–14.6)	0.88

^aClinical T-staging not applicable for 1 patient

^bMesorectal N Category; Clinical N-staging not applicable for 1 patient

^cMRF: mesorectal fascia; Involvement of MRF not applicable for 1 patient

Surgical Treatment and Perioperative details:

Surgical treatment and perioperative details are summarized in Table 2. The majority of patients underwent low or ultra-low anterior resection (n=91, 57.6%) and the sphincter preservation rates did not differ (p=0.23) between groups. There was a numerically greater frequency of multi-visceral resection surgery in the LPLND (12.5% vs. 4.3%; p=0.09) group and more patients in the LPLND group (47.7%) underwent minimally-invasive surgery than the non-LPLND group (30.0%; p=0.02). The median operative time was greater with LPLND (509 minutes vs. 429 minutes; p=0.003), but there was no difference in the median blood loss (350 ml vs. 400 ml; p=0.70) (Table 2).

Table 2:

Surgical Treatment and Perioperative Outcomes for complete cohort of rectal cancer patients with pre-treatment abnormal lateral pelvic lymph nodes undergoing TNT followed by either lateral pelvic lymph node dissection at the time of total mesorectal excision (TME) or TME only

	LPLN Non-Dissection (n = 70)	LPLN Dissection (n = 88)	p-value
ASA-classification (%)			0.51
2	2 (2.9)	5 (5.7)
3	66 (94.3)	82 (93.2)
4	2 (2.9)	1 (1.1)
BMI (kg/m2) (median, IQR)	29.5 (26–35)	28.4 (25.3–32.5)	0.23
Approach (%)			0.02
Open	49 (70.0)	46 (52.3)
Robotic	21 (30.0)	42 (47.7)
Type of resection (%)
Sphincter-Preserving	44 (62.9)	47 (53.4)	0.23
Exenteration	3 (4.3)	11 (12.5)	0.09
Lymphadenectomy Laterality			N/A
Bilateral	N/A	17 (19.3%)
Unilateral	N/A	71 (80.7%)
Operation time (minutes) (median, IQR)	429 (354–529)	509 (406–644)	0.003
Estimated Blood Loss (mL) (median, IQR)	400 (150–600)	350 (150–1000)	0.70
Blood Transfusions (patients) (%)	13 (18.6)	20 (22.7)	0.52
Length of hospital stay (days) (median, IQR)	5.0 (4.0–6.0)	6.0 (4.0–8.0)	0.12
Clavien-Dindo Major Complication (%)			0.73
Grade 3+ (%)	12 (17.2)	17 (19.3)	0.73
Unplanned readmissions within 30 days (%)	11 (15.7)	14 (15.9)	0.97
Reoperation within 30 days (%)	1 (1.4)	3 (3.4)	0.63

Postoperative outcomes:

Short-term perioperative outcomes were examined (Table 2). There was no significant difference in median length of stay (p=0.12), or rates of major morbidity (19.3% vs. 17.0%; p=0.73), or 60-day mortality (p=1.00) between groups. Additionally, rates of reoperation (3.4% vs. 1.4%; p=0.63) or readmission (15.9% vs. 15.7%; p=0.97) were not associated with performance of LPLND. (Detailed information regarding specific complications stratified by sphincter preservation are described in Supplementary Table 1 and Supplementary Figure 1). However, there was a higher rate of transient urinary retention in the LPLND group (22.7% vs. 4.3%; p<0.001).

Oncological outcomes:

Details regarding pathological outcomes are shown in Table 3. Whilst there was a higher rate of lymphovascular invasion (44.3% vs. 34.3%; p=0.02), the rate of perineural invasion was lower (31.8% vs. 41.4%; p=0.05) among patients who underwent LPLND when compared to non-LPLND patients. Resection of the rectal cancer and concurrent LPLND occurred in 85 (96.6%) patients, whilst the remaining 3 (3.4%) patients had subsequent LPLND after a prior rectal cancer resection. The median lymph node count from the lateral compartment dissection was 5 (IQR 2–8), with 30 (34.1%) patients having one or more positive LPLNs.

Table 3:

Pathological data for resection of the primary rectal cancer with or without concurrent lateral pelvic lymph node dissection.

	LPLN Non-Dissection (n = 70)	LPLN Dissection (n = 88)	p-value
pT-stagea (%)			0.82
pT0	4 (5.7)	6 (6.8)
pT1	6 (8.6)	8 (9.1)
pT2	10 (14.3)	13 (14.8)
pT3	40 (57.1)	47 (53.4)
pT4	10 (14.3)	11 (12.5)
pN-stage (%)			0.17
pN0	43 (61.4)	42 (47.7)
pN1	20 (28.6)	28 (31.8)
pN2	7 (10.0)	15 (17.1)
Margins + (%)	1 (1.4)	5 (5.7)	0.23
LVI+ (%)	24 (34.3)	39 (44.3)	0.02
PNI+ (%)	29 (41.4)	28 (31.8)	0.05
Differentiationc (%)			0.36
Well differentiated	2 (2.9)	7 (8.0)
Moderately differentiated	50 (71.4)	63 (71.6)
Poorly differentiated	18 (25.7)	18 (20.5)
TDs (%)	17 (24.3)	21 (23.9)	0.95
TRGd (%)			0.60
0	3 (4.3)	8 (9.1)
1	15 (21.4)	15 (17.1)
2	43 (61.4)	52 (59.1)
3	9 (12.9)	11 (12.5)
Number of patients with positive LPLNs (%)		30 (34.1)	N/A
Number of LPLN Harvested, (median, IQR)		5 (2–8)	N/A

LVI, lymphovascular invasion; PNI, Perineural invasion; TDs, Tumor Deposits, TRG, Tumor Regression Grade

^apT-staging not applicable for 3 patients

^bpN-staging not applicable for 3 patients

^cLymphovascular and Perineural invasion not applicable for 6 patients

^dTumor Regression Grade not applicable for 3 patients

No differences in the rates of recurrence were observed between groups. The 3-year lateral local recurrence rates were 3.4% for LPLND and 4.6% for non-LPLND (p=0.85). The 3-year overall recurrence rates were 33.9% versus 36.7% (p=0.74), local recurrence rates were 12.2% versus 9.6% (p=0.44) and distant recurrence rates were 24.9% versus 28.2% (p=0.63) for the LPLND and non-LPLND groups, respectively (Figure 1a–d). Overall (p=0.99) and disease-free survival (p=0.93) also did not differ between the LPLND and non-LPLND groups. However, after stratifying the LPLND group based on pathologic LPLN status, DFS rates were higher in the non-LPLND and the LPLND with pathologically negative LPLN groups, when compared to the LPLND group with pathologically positive LPLNs (p=0.02; Figure 2a–c).

Figure 1:

Cumulative risk of recurrence by treatment group: 1a. Lateral local recurrence; 1b. Overall recurrence; 1c. Distant recurrence; 1d. Local recurrence.

Figure 2:

Comparison of survival outcomes: 2a. Overall survival; 2b. Disease free survival; 2c. Disease free survival by LPLND and pathological lateral pelvic lymph node metastasis status.

MRI Criteria for LPLND

All patients were identified to have enlarged LPLN during initial workup, but both the pre-TNT and post-TNT LPLN sizes were greater in the LPLND versus non-LPLND groups (Table 4). The median pre-TNT lymph node size (short axis) was 10 mm (IQR 7–14) for the LPLND versus 6.0 mm (IQR 5–8) for the non-LPLND group and the median post-TNT lymph node size was 6.0 mm (IQR 4–8) for the LPLND versus 4.0 mm (IQR 0–5) for the non-LPLND groups (p<0.0001). Other malignant features (e.g. heterogeneous signal or irregular margin) were more prominent on the pre-TNT MRI for the LPLND group compared to the non-LPLND group (67.1% vs. 38.6%; p=0.004). There was a similar frequency in the site of the pre-TNT enlarged LPLN (p=0.26).

Table 4:

MRI characteristics of lateral pelvic lymph nodes

	LPLN Non-Dissection (n = 70)	LPLN Dissection (n = 88)	p-value
SA Sizea (mm) (median, IQR)
Before TNT	6.0 (5.0–8.0)	10.0 (7.0–14.0)	0.0001
After TNT	4.0 (0–5.0)	6.0 (4.0–8.0)	0.0001
Site of LPLN (%)			0.26
Internal Iliac	35 (50.0)	39 (44.3)
Obturator	34 (48.6)	43 (48.9)
Common Iliac	1 (1.4)	6 (6.8)
Malignant features (%)
Pre-TNT (%)b			0.004
Enlarged, no adverse features	38 (54.3)	25 (28.4)
Heterogenous Signal	8 (11.4)	9 (10.2)
Irregular Margin	6 (8.6)	18 (20.5)
Both	13 (18.6)	32 (36.4)
Post-TNT (%)c			0.07
Enlarged, no adverse features	42 (60)	39 (44.3)
Heterogenous Signal	8 (11.4)	12 (13.6)
Irregular Margin	6 (8.6)	18 (20.5)
Both	7 (10.0)	15 (17.1)

^aSA: Short-axis LPLN size

^bMissing 9 pre-TNT MRI for 9 patients

^cMissing 11 post-TNT MRI for 11 patients

The factors associated with receipt of LPLND were identified and incorporated into a multivariable logistic regression model (Figure 3), comparing different categories of the lymph node characteristics on pre- and post-TNT MRI. The decision for LPLND was associated with either pre-TNT abnormal MRI features or post-TNT enlarged size. Patients with pre-TNT LPLNs demonstrating heterogeneous signal and/or irregular margin (OR 3.82; 95% CIs: 1.65–8.82) and patients with a short axis LPLN ≥ 5 mm post-TNT (OR 2.69; 95% CIs: 1.19–6.08) were more likely to undergo LPLND.

Figure 3:

Multivariable analysis for factors affecting decision making for LPLND versus non-LPLND.

DISCUSSION:

A consensus approach for the assessment and treatment of LPLNs is still evolving and the additional value of therapeutic LPLND in reducing lateral local recurrence risk is currently a hot topic of discussion in the multidisciplinary community treating rectal cancer(4). This series demonstrates that among rectal cancer patients with pre-TNT enlarged LPLNs, the appropriate selection of patients for LPLND after TNT can be facilitated by a multidisciplinary MRI directed approach without increasing the risk for lateral local recurrence rates. Furthermore, there was no difference in peri-operative or oncological outcomes between LPLND and non-LPLND patients. This study demonstrates that MRI directed multidisciplinary decision-making strategy that combines morphologic features and LPLN size can facilitate the selection of patients for LPLND following TNT. Furthermore, the similar rates of lateral local recurrence between the two groups, suggests those that underwent LPLND with pathologically positive LPLNs, were spared the potential lateral local recurrence.

The previous equipoise between Eastern and Western strategies for the management of LPLN metastases demonstrated that not all patients required LPLND nor was neoadjuvant chemoradiation sufficient to prevent lateral local recurrence either(37). Recent studies have shown that up to two-thirds of patients suspected to have LPLN metastases and treated with neoadjuvant chemoradiation and selective LPLND based on pre- and post-treatment LPLN size harbored pathologically confirmed LPLN metastases(13). However, there is currently no consensus method for predicting patients with LPLN metastases following neoadjuvant chemoradiation and the traditional strategy has been based on the size of the LPLN on imaging. The current NCCN guidelines advocate against LPLND unless clinically suspicious for metastasis(23), and there are associated risks to LPLND, therefore, it is imperative that decision-making treatment strategies are developed to accurately identify which patients are most likely to have LPLN metastases following neoadjuvant therapy. In this study, the overall rate of LPLN positivity was 34.1%. It can be argued that while a very low rate of positivity suggests that the criteria for selection are too permissive, a higher rate may indicate that the criteria are too stringent, resulting in a greater risk of non-dissection of residual disease and disease recurrence. Thus, the optimal balance between sensitivity and specificity requires further investigation.

Traditionally, LPLND has rarely been performed in Western countries, thus it has been difficult to gather sufficient data to facilitate a decision-making strategy for identifying patients who would benefit from LPLND. However, studies have shown that neoadjuvant chemoradiation alone for patients with enlarged LPLNs was not sufficient for local control(38–40). A recent international multi-center cohort study analyzed outcomes of multidisciplinary treatment with or without LPLND for patients with enlarged LPLNs and showed a benefit to LPLND, however, more than 90% of the patients treated with LPLND were from Japan(12). There is also a lack of consensus regarding whether LPLND is still necessary for patients with LPLNs demonstrating treatment response on restaging MRI. Moreover, the literature mainly pertains to studies from Eastern centers where LPLND is performed(41). Thus, it is unclear whether leaving nodes worsens the lateral local recurrence rate(24). A recent multi-center comparative study assessed the local recurrence rates in Western patients with low rectal cancer and suspicious LPLN treated with or without LPLND after neoadjuvant chemoradiation(42). This study demonstrated that LPLND could safely be performed in Western patients treated with neoadjuvant chemoradiation, and that LPLND was associated with improved locoregional control(42). However, a limitation of this study was the heterogeneity between the different centers, including treatment strategies, surgical technique and patient populations(42).

Traditionally, size of the LPLN has been the main criteria to determine the need for therapeutic LPLND(13). A previous study reported that LPLND could potentially be avoided if the short axis LPLN diameter can be reduced from ≥ 7 mm to ≤ 4mm on MRI following neoadjuvant chemoradiation(24). A more recent study has shown that LPLN metastases can still be detected in cases with a short axis LPLN diameter ≤ 4 mm despite neoadjuvant chemoradiation(43). Despite the differences in imaging modality to assess the LPLN size, which might explain the discrepancy, this study highlights the limitations of size as the single criterion for therapeutic LPLND(43). Other malignant features, such as internal heterogeneity and/or border irregularity, in pre-treatment enlarged LPLNs may also be predictive of oncological outcomes, as enlarged LPLNs harboring malignant features may be associated with worse distant metastasis free survival(44). This present study demonstrates that in addition to size, pre-TNT malignant features in LPLNs were an independent predictor for LPLND. These malignant features in LPLNs appear to be an important independent predictor for LPLN metastases, and therefore should be considered in addition to size in treatment stratification(45). As demonstrated by the pathological data, the MRI criteria for LPLND were also associated with the pathological identification of high-risk features such as lymphovascular or perineural invasion, highlighting the reliability of MRI criteria for identifying high-risk patients most likely to benefit from LPLND. Radiomics-based prediction modeling has also been shown to outperform the standard pre-treatment LPLN short-axis diameter, and may further provide an individualized risk estimation for LPLN metastases in rectal cancer patients(43).

Previous studies have shown that a pathological complete response can be achieved in up to 20% of patients after conventional neoadjuvant long-course chemoradiotherapy(46, 47). The NCCN guidelines now support the use of TNT in the management of locally advanced rectal cancers(23), and recent studies have demonstrated a sustained clinical or pathological complete response in over 30% of patients after TNT(48, 49). It is therefore imperative to establish models that help predict LPLN metastases in patients treated with TNT. All previous studies on LPLNs have evaluated the role of upfront surgery or long-course chemoradiation followed by re-evaluation for LPLND. This is the first study to assess the role of LPLND or observation in patients presenting with evidence of LPLN metastases prior to completing TNT and then to also evaluate the multidisciplinary decision-making algorithm to determine which patients will most likely benefit from LPLND.

There has been a general reluctance from Western surgeons to perform LPLND in the past due to the associated challenges of performing the procedure, potential risk of intraoperative bleeding and the associated post-operative morbidity(50). However, the current study did not show any differences in the median blood loss, median length of stay or major morbidity between the LPLND and the non-LPLND groups. Thus, this study demonstrates that LPLND for locally advanced rectal cancer can be performed in Western patients following TNT to completely resect extra-mesorectal LPLNs and that it is associated with acceptable peri-operative morbidity and short-term oncological outcomes.

There are several notable limitations to this study. This is a retrospective study and subject to associated bias, however, it was based on data collected prospectively, including the data regarding complications. Although data regarding pathologic and perioperative outcomes, including perioperative urinary dysfunction, were prospectively collected, long-term functional outcomes, such as sexual dysfunction, were assessed retrospectively. The low figure of reported sexual dysfunction therefore may highlight a limitation of the retrospective assessment based on provider or patient solicitation during a clinic visit and is thus subject to recall bias. The relative short follow-up interval is also an important limitation. However, there was no observed difference in the low recurrence rates in both groups, and given that the majority of recurrences will occur within the first 3 years, significant change in this finding with longer follow up is not anticipated. The data were obtained from a single-centre and the surgical interventions were heterogeneous, thus the findings may not be generalisable to all units, particularly those that do not have experience in performing LPLND. Conversely, this series reflects the outcomes of a standardised approach to the management and surgery of patients with clinical evidence of LPLN metastasis. The effect of change in size of the LPLN and other pathological or tumor traits were explored. Post-treatment lymph node size was associated with pathological positivity, however, but there was insufficient data for specific quantitative analysis of size change and pathological status. Therefore, this is certainly an area for future work. However, in the absence of prospective studies, this report adds further data to the paucity of literature on outcomes following combined modality treatment with LPLND versus non-LPLND in Western patients.

CONCLUSION:

For rectal cancer patients with evidence of LPLN metastasis, the appropriate selection of patients for LPLND can be facilitated by pre- and post-TNT multidisciplinary MRI directed approach with no significant difference in perioperative or oncologic outcomes.

Paper24. Magnetic Resonance Imaging (MRI) Directed Surgical Decision Making for Lateral Pelvic Lymph Node Dissection (LPLND) In Rectal Cancer after Total Neoadjuvant Therapy (TNT)

*Oliver Peacock¹, *Naveen Manisundaram¹, *Youngwan Kim¹, *Nir Stanietzky², *Emma B. Holliday³, *Arvind Dasari⁴, *Melissa Taggart⁵, *Harmeet Kaur², *Brian Bednarski¹, *Tsuyoshi Konishi¹, George J. Chang¹

¹Colorectal Surgical Oncology, The University of Texas MD Anderson Cancer Center, Houston, TX, ²Radiology, The University of Texas MD Anderson Cancer Center, Houston, TX, ³Radiation Oncology, The University of Texas MD Anderson Cancer Center, Houston, TX, ⁴Medical Oncology, The University of Texas MD Anderson Cancer Center, Houston, TX, ⁵Pathology, The University of Texas MD Anderson Cancer Center, Houston, TX

Dr. Matthew Kalady (Columbus, OH):

Good afternoon, congratulations to Dr. Peacock and the multidisciplinary group from M.D. Anderson on an excellent paper. For rectal cancer patients accurate clinical diagnosis of metastatic spread to lateral pelvic lymph nodes remains a challenge and the most appropriate management of these nodes is still controversial. As we heard, without proper treatment these nodes are an important source of local recurrence and morbidity. There are primarily two different schools of thoughts regarding management. One is routine lateral pelvic lymph node dissection which is championed in the eastern part of the world, mainly Japan. And the other is chemoradiation treatment to the nodal basins which is the primary approach in the United States followed by lymph node dissection in select patients. Lateral pelvic lymph node dissection is a technically challenging procedure, requires specific training and expertise, and even in skilled hands adds time and morbidity to the case. Who should undergo lymph node dissection is not well-defined and there is a paucity of data comparing the outcomes of these different approaches. Furthermore, the field is moving towards total neoadjuvant therapy which we know has better rates of sterilization against mesorectal nodes and adding consolidation chemotherapy in this approach might also provide better local control of the lateral nodes and limit the need for lateral node dissection. This is clearly an area where more information is needed to help guide management. In the study presented today, all 158 patients that had clinically suspicious lateral pelvic nodes received total neoadjuvant therapy and about half of those underwent lateral pelvic lymph node dissection, along with a TME. The authors report that using their patient selection approach they can achieve similar oncologic outcomes between patients who they deemed needed or did not need a lateral lymph node dissection. Note though that about two thirds of the patients who had dissection did not have tumor in those nodes. The lymph node dissection did come at a cost of higher rates of urinary tension and ileus. It is important that we know this information when we discuss outcomes and expectations with our patients. I would also caution putting too much emphasis on the report of similar oncologic outcomes given that the study has not shown a difference but the median follow- up time is relatively short at 20 months, and 25% of patients having less than 10-month follow-up. It will be interesting to see these results in a longer term if you have the ability and plans to do so. One potential surrogate for oncologic outcome though that is readily available is the detection of pathological positive lymph nodes. The authors present factors that were associated with the choice to do the nodal dissection in a retrospective analysis and that choice seems to have resulted in finding pathologic lymph nodes in 34% of cases. Thus, about two-thirds of patients may have had an unnecessary and morbid procedure. In my opinion, the positive cases are the cases that are most relevant. So, my first question is are you able to analyze the data for the pathologically positive cases so that you can predict up front who would have the positive nodes? And then secondly, related to this were you able to find any common characteristics for the group that was dissected or had negative nodes and thus maybe be able to come up with a more personalized approach to determine who might benefit most while avoiding the morbidity in those that could not? Thank you Dr. Sarr and the program committee for the opportunity to discuss this presentation. Again congratulations on this very important work.

Response from Oliver Peacock:

We would like to thank Dr. Kalady for his expertise in reviewing this manuscript and for providing a very nice summary of the paper. There was no significant difference in the major morbidity between the two groups when comparing LPLND versus non-dissection, however, as Dr. Kalady has highlighted, the study did show that there was a higher rate of transient urinary retention following LPLND which was 22.7% of patients. However, unpublished data from our group has shown there is a reduction in the rate of urinary retention with increasing surgeon experience of performing LPLND. LPLND is also associated with an increase in operating time. Previously the other major concerns or barriers to performing this procedure in the past have been related to the risk of blood loss or technical feasibility in Western patients, but this study demonstrates that it can be performed in this cohort of patients with no difference in blood loss.

We agree with Dr. Kalady that one of the limitations of this study is the relatively short median follow-up of 20 months. However, there was no observed difference in the low recurrence rates in both groups, and given that the majority of recurrences will occur within the first 3 years, significant change in this finding with longer follow up is not anticipated. Moreover, we do plan to continue to follow these patients for a longer period of time to assess whether there is evidence of changes in the overall survival outcomes and recurrence rates for these patients.

Dr. Kalady also highlights a great area of research interest, with regards to analyzing those patients that had pathologically positive lateral pelvic lymph nodes because 30 patients out of the 88 that were dissected had an actual positive lymph node. We currently do not know what the correct number of pathologically positive patients should be after LPLND, but based on the current limited evidence and international experience, 34% positivity is a fair yield. Therefore, we plan to continue reviewing our decision making strategies and further investigate features to further facilitate the identification of patients most likely to benefit from LPLND, and avoid this procedure and associated risks in others. The take home message is that currently this MRI-directed approach provides a suitable strategy for identifying patients that require LPLND, but further work is required in order to more selectively dissect patients and avoid the associated risks of the procedure in others. Once again, we would like to take the opportunity to thank Dr. Kalady for reviewing this manuscript and for providing expert feedback.

Dr. Jose Guillem (Chapel Hill, NC):

Jose Guillem from Chapel Hill. I congratulate you and Dr. Chang for the excellent work and presentation. I agree with Dr. Kalady and believe that two years is a rather short follow-up time period to be able to comment on recurrence. I would also like you to comment that your reported sexual dysfunction rate of single digits is remarkable especially when you compare to the sexual dysfunction rate from Sloan Kettering reported by Enker, Cohen et al about 20–30 years ago. Are the differences due to differences in the age of the population or usage MIS/Robotic platform?

Response from Oliver Peacock

We would like to thank Dr. Guillem for his comments and question. We think the low figure of reported sexual dysfunction highlights a limitation of the retrospective nature of our data and agree that it is subject to recall bias. Moreover, sexual function was not prospectively evaluated, thus was subject to provider or patient solicitation during a clinic visit. However, there were a relatively high number of patients that underwent minimally invasive LPLND in this cohort, which does help to facilitate identification of the nerves and neurovascular bundles, which might also contribute to a lower rate of sexual dysfunction. Future work will include the use of validated patient reporting outcome measures and collection of this data prospectively, in order to gather the true rate of sexual dysfunction.

Dr. David Shibata (Memphis, TN):

Dave Shibata from Memphis I also offer my congratulations to you and Dr. Chang and the rest of your group for a really important study. One question I had for you is related to your selection criteria. It was not explicitly stated but did you use the Lateral Lymph. Node Consortium guidelines for post-treatment size that suggest a no greater than 4mm cut off. I believe I saw that your upper range was 5mm for your non-pelvic lymph node dissection patients. My second question to you is related to the next step beyond TNT, which is our growing interest in “watch and wait”. Could you comment on the implications of such studies on that area of work and also whether there any excluded patients from the study that perhaps fell into that category of care? Thank you very much.

Response from Oliver Peacock:

We would like to thank Dr. Shibata for those questions. So, for the benefit of the audience, the lateral node consortium is a multi-center collaborative that has done some excellent work at trying to create a consensus for LPLND. A limitation associated with this collaboration in addition to its retrospective nature, was that the majority of the Western patients underwent chemoradiation without dissection while the majority of patients who underwent LPLND after chemoradiation were from Eastern centers. The consortium uses size of the LPLN as the main criteria, with the cut-off being ≥ 7 mm on pretreatment MRI and ≤ 4 mm on post treatment MRI.

For the present study, we did not exclusively use the size criteria from the consortium. Size of the LPLN is an important factor, and our study demonstrates that ≥ 5 mm LPLN after TNT was an independent predictor for LPLND. However, there are other features associated with LPLNs other than size, and therefore the aim of this study was to build on the work of the collaborative, and investigate other potential factors that might help to determine the decision making for LPLND. We showed that pre-treatment morphological features were an important predictor. When it comes to the multidisciplinary review, we may not necessarily dissect a patient with a post-treatment LPLN of 5 mm, if all the other features look relatively benign and also the change in the size was not particularly significant either.

In regards to watch and wait in the era of total neoadjuvant therapy, previously after long course chemoradiation, we have observed that approximately 20% of patients will have a complete pathological response. Following total neoadjuvant therapy, we are seeing reports that as much as 40 – 50 % of patients can achieve a clinical complete response and be considered for watch and wait. Some of the eligibility for watch and wait is very dependent on the initial presentation of the disease. We do have a few patients at our center who had initially clinically positive LPLN but based on response have undergone watch and wait, some cases after LPLND; but unfortunately, we do not have enough data yet to make further comment. It is however, an interesting area for future work.

Dr. David Shibata (Memphis, TN):

One more quick question, did you look at the change, the effect of change in lymph node size in terms of positive nodes? You separated these preoperatively by their pre-PTMT size, how about the change because there has been some work in looking at changes?

Response from Oliver Peacock

Thank you, that is a great question. We did explore the effect of change in size of the LPLN. Post-treatment lymph node size was associated with pathological positivity, however but there was insufficient data for specific quantitative analysis of size change and pathological status. This is also certainly an area for future work.

Dr. Michael Sarr:

It would be interesting to correlate that with total pathologic features.

Response from Oliver Peacock:

Yes, we would agree with that statement and may also facilitate the identification of those patients that underwent LPLND and were pathologically negative.