Technique to match mesorectal lymph nodes imaging findings to histopathology: node-by-node comparison

Abstract

Background

Lymph node status is critical for staging rectal cancer and determining neoadjuvant therapy regimens. Establishing a matching between imaging and histopathological lymph nodes is fundamental for predicting lymph node status. This study reports a technique to achieve node-by-node pairing of mesorectal lymph nodes between imaging findings and histopathology.

Methods

Fifty-two patients with histopathologically verified rectal cancer underwent magnetic resonance imaging before surgery. The status of each lymph node in the surgical specimens was analyzed histopathologically and matched with preoperative imaging after the operation.

Results

A total of 346 mesorectal lymph nodes were located on imaging evaluation, of which 313 were confirmed histopathologically, and 33 were unmatched. The total success rate of the technique was 90.5%. Node-by-node analysis revealed 280 benign and 33 malignant nodal structures.

Conclusion

The technique to match mesorectal lymph node imaging findings to histopathology was feasible and effective. It simplified the technical method and had a reasonable success matching rate, which could provide a standardized approach for obtaining a prospective correlation between imaging and histological findings, supporting all subsequent related studies at the level of mesorectal lymph nodes.

Introduction

Evaluation of lymph node (LN) status is crucial for deciding the treatment course of rectal cancer. Patients suspected of having local LN metastasis and recurrence should receive adjuvant therapy. Although the previous risk assessment was mainly based on postoperative histopathology, adjuvant therapy has gradually shifted from a postoperative to preoperative focus in the past few decades (Kapiteijn et al. 2001; Habr-Gama et al. 2006). In the context of neoadjuvant therapy, preoperative imaging must accurately diagnose regional LNs; however, since no uniform standard for defining LN status imaging exists, conventional imaging methods are less reliable in detecting LN metastases (Bipat et al. 2004; Zhuang et al. 2021; Li et al. 2015). Simultaneously, the current N staging mainly relied on the number of positive LNs found in the mesorectum after the overall sampling of rectal specimens and using histological findings to assess the patient's LN status. This method did not involve directly assessing LNs on imaging studies, and, as a result, had questionable accuracy.

Several studies have begun exploring new methods to improve N staging of rectal cancer in recent years, including lymphatic contrast agents, chemical shift effects, and radiomics models (Zhang et al. 2017; Harisinghani et al. 1998; Liu et al. 2021). Notably all these studies required the extraction of imaging features of individual LNs. However, the significant difficulty of using LN features is the lack of pathological ground truth in each LN detected in the images. Thus, matching individual LNs seen on images to their precise histological counterparts and achieving a node-by-node correspondence between imaging and histopathology is undoubtedly an urgently needed pre-step for prospective studies (Brown et al. 2003; Koh et al. 2004, 2005; Park et al. 2014; Rutegård et al. 2020). However, obtaining an accurate histology-imaging correlation is technically challenging.

Through this article, we propose a technique to achieve node-by-node pairing of mesorectal LNs between imaging findings and histopathology. Preliminary results have verified the practicality and feasibility of this method. To the best of our knowledge, this is the first report to describe this type of technology in detail.

Materials and methods

This methodological study was based on a prospective study for imaging accuracy in diagnosing mesorectal lymph node staging (ChiCTR2100052441). The Medical Ethics Committee of West China Hospital approved the study, and written informed consent was obtained from all patients before surgery. From July 2021 to May 2022, patients with histopathologically verified rectal cancer underwent rectal MRI before surgery. The specific inclusion criteria were: (1) rectal carcinoma located ≤ 10 cm above the anal verge; (2) radical surgery scheduled within two weeks after rectal MRI; and (3) patients with 1 ≤ LNs ≤ 15 on preoperative imaging findings.

MR imaging protocol

MR imaging was performed with a 3 T Magnetom Skyra MR scanner (Siemens Healthineers, Malvern, PA, USA) employing an 18-channel body coil; All patients were given an intravenous antiperistaltic agent (10 mg raniscopolamine hydrochloride) 30 min before MRI for bowel preparation. High-resolution rectal MRI protocol comprised turbo spin-echo sagittal, oblique coronal, oblique axial T2- and diffusion-weighted imaging. The scan parameters used for the oblique axial T2-weighted imaging sequence were as follows: repetition time/echo time, 6,890/100; slice thickness, 3 mm; voxel size, 0.3 × 0.3 × 3 mm; field of view, 180 mm; matrix, 384 × 346; slices, 48; average, 3; total scanning time, 5 min and 5 s; and parallel acquisition technique with generalized auto-calibrating partial parallel acquisition acceleration factor. Oblique axial DWI sequence was a transverse echo-planar imaging diffusion sequence with 1000 s/mm² as the highest b value. The same parameters such as field of view (FOV), slice thickness and gap were used in DWI to match the tumor on the oblique axial T2WI. The total scan time was 30 min.

Technical notes

The technique is described in detail in the following steps:

Imaging evaluation

All rectal magnetic resonance imaging (MRI) of patients with rectal cancer at the Department of Radiology were analyzed preoperatively by a radiologist with more than 15 years of experience in the interpretation of rectal imaging studies as well as two gastrointestinal surgeons. MRI high-resolution oblique-axis T2-weighted imaging and DW images were used as evaluation sequences. LNs on DWI were defined as a round or oval structure showing high signal intensity. After referencing the DWI, the radiologist performed anatomic correlation matching with T2WI to confirm regional LNs and record the following characteristics:

1. Nodal size: the largest short-axis diameter of each LN.

2. Nodal location: (a) Vertical positioning: Using the sagittal and axial images to estimate the distance of each node from the inferior tumor margin, approximated to the nearest axial slice. The cranial-caudal distribution of the mesorectal nodes in relation to the primary tumor was evaluated. The number of mesenteric LNs found at the tumor level, above the proximal tumor margin or below the distal tumor margin were also noted. (b) Horizontal positioning: the minimum nodal distance to the mesorectal fascia and rectal wall was measured by axial images, and the positional relationship of each node to surrounding blood vessels and adjacent nodes was recorded.

3. Level sequence of each LN. Only the LNs with matched images on T2WI and DWI were evaluated (excluding the lateral pelvic LNs). After initially judging the abovementioned indicators, will then be referred for senior radiologist review to make the final judgment (Fig. 1).

Fig. 1.

Lymph nodes matched by MRI images. A In rectal MRI-T2WI image. B In rectal MRI-DWI image. All images show same 4 × 5 mm lymph node (left arrow) and 3.5 × 5 mm lymph node (right arrow) within mesorectum

Anatomical maps

All LNs evaluated by imaging were drawn on anatomical maps to obtain accurate histology-imaging correlation node-by-node. The radiologist and two gastrointestinal surgeons who participated in the imaging evaluation drew horizontal and vertical anatomical maps. The layer sequence as well as the distance of each node from the inferior tumor margin were marked on the vertical maps. The size and shape of each node, the minimum nodal distance to the mesorectal fascia, rectal wall, and the positional relationship with the surrounding tissues were marked on the horizontal maps (Fig. 2).

Fig. 2.

A Easy-to-use standardized template. Vertical map: estimate the distance of each node from the inferior tumor margin (point 0), numbering sequentially from the starting point to proximal, the actual distance corresponding to adjacent numbers is 3 mm. Evaluate the cranial-caudal distribution of the mesorectal nodes in relation to the primary tumor and record the number of LNs found at each slice. Horizontal maps: mark the size and shape of each node at the corresponding slices, the minimum nodal distance to the mesorectal fascia, rectal wall, and the positional relationship with the surrounding vessels. B Anatomical maps of a 55-year-old male with rectal cancer

Surgical specimen processing

All patients underwent total mesorectal excision (TME) within 2 weeks of the imaging examination. The operations were performed by an experienced laparoscopic colorectal surgeon (ZQ-W). The fresh specimens were opened the above upper border of the mesorectum in the operating room (at least 2 cm above the tumor), and marked the midline of the mesorectal long axis. The specimens were then pinned to a foam board and sent for histopathological analysis (Fig. 3).

Fig. 3.

Surgical specimen processing. A Marked the midline of the mesorectal long axis on fresh specimens. B Specimen's proximal, middle, and distal ends were pinned to foam board and fully straightened under tension

Nodal matching and radiologic-pathologic comparison

The pathology laboratory conducted a macroscopic examination of the fresh surgical specimens after they arrived, and then immersed them in formalin saline for at least 72 h for fixation. The pathological evaluation was completed by a pathologist with 15 years of experience, the aforementioned experienced radiologist, and the two experienced gastrointestinal surgeons mentioned prior. First, each specimen was guided by a 3 mm ruled template and sectioned at 3 mm intervals transversely, perpendicular to the long axis of the mesorectum, from the distal aspect to the proximal aspect. All the slices were numbered and photographed (Fig. 4). We carefully identified LNs on each slice and used the anatomical map as a template for node-by-node correspondence. The precise matching process mainly depended on the size and location (orientation, distance to rectal wall, or mesorectal fascia) of the LNs. All matched LNs were numbered and placed in individual trays for processing. In subsequent microscopy, all slides were stained with hematoxylin and eosin. The benign and malignant LNs were reported according to the results of microscopy (Fig. 5).

Fig. 4.

Surgical specimen sliced and photographed. A Specimen was sectioned transversely stepwise from distal to proximal ends and photographed

Fig. 5.

Nodal matching among in rectal MRI and tissue slice (arrow). A In rectal MRI-T2WI image. B In rectal MRI-DWI image. C Transverse tissue slice. D Nodes harvested from each tissue were placed in individual trays. Nodal matching was performed by relating the position of each node (circled) to the position of tumor/rectum, and by comparing the location of the node to blood vessels within the mesorectum (arrows)

Clinical and imaging data collection

The clinical data included patient age, gender, carcinoembryonic antigen (CEA) and carbohydrate antigen 19-9 (CA19-9) levels. The histological grades were obtained from pathological reports.

Lymph node-related imaging indicators were measured at the axial T2-weighted imaging (T2WI), including the maximum short diameter (Sd, mm), the minimum nodal distance to the mesorectal fascia and rectal wall (mm), and the number of LNs found at the tumor level, above the proximal tumor margin or below the distal tumor margin.

Result

From July 2021 to May 2022, fifty-two initial diagnosed patients who did not receive preoperative neoadjuvant therapy were prospectively enrolled. The patients’ basic information is presented in Table 1.

Table 1.

Clinical features of patients

Clinicopathologic feature	Number of patients
Age, median (range)	67 (37–88)
Gender
Male	23 (44.2%)
Female	29 (55.8%)
Location
Low (< 5 cm)	14 (26.9%)
Middle (5–10 cm)	38 (73.1%)
pT-stage
T1	2 (3.8%)
T2	12 (23.1%)
T3	35 (67.3%)
T4	3 (5.8%)
pN-stage
N0	35 (67.3%)
N1	14 (26.9%)
N2	3 (5.8%)
CA19-9/(U mL−1)
0–27	46 (88.5%)
> 27	6 (11.5%)
CEA/(ng mL−1)
0–5	34 (65.4%)
> 5	18 (34.6%)

pT-stage pathological T stage, pN-stage pathological N stage, CA19-9 carbohydrate antigen 19-9, CEA carcinoembryonic antigen

For evaluation on a node-by-node basis, only those LNs that could be identified were included. A total of 346 LNs were identified during imaging evaluation. The mean number of nodes identified per patient was 6 (range 3–12). Of these, 313 were confirmed on histopathological findings, and 33 were unmatched. The overall success rate of the technique was 90.5%. Figure 6 summarizes the histologic findings of 660 lymph nodes in 52 rectal specimens.

Fig. 6.

Histologic findings of 660 lymph nodes in 52 rectal specimens

Node-by-node analysis showed that of the 313 matched nodes, 280 (89.5%) LNs were negative and 33 (10.5%) were positive. Of the 347 unmatched nodes found on histopathological examination, 330 were negative and 17 were positive. The incidence of positive nodes found on histopathological examination was 50/660 (7.6%).

Nodal size

The MRI median short-axis diameters were 4.0 mm (range 2.0–9.0 mm), 5.6 mm (range 2.5–11.8 mm) for the negative and positive nodules, respectively.

Nodal distribution

Distance to the distal tumor margin

In all matched nodes, 146/313 (46.6%) were found at the level of the tumor; Ninety-eight percent (306/313) of nodes were seen at the level of the tumor or within 5 cm proximal to the tumor. Only 5 nodes were visualized below the distal tumor margin. All positive LNs were distributed at the level of the tumor or within 5 cm proximal to the tumor.

Minimum nodal distance to mesorectal fascia and rectal wall

The mean minimum nodal distance to the rectal wall and mesorectal fascia was 9.0/4.3 mm, 5.9/5.7 mm for the negative and positive nodules (Table 2).

Table 2.

Imaging characteristics versus histologic findings in 313 matched nodal

Imaging characteristics	Histologic findings
	Benign	Malignant
Matched nodes	280 (89.5%)	33 (10.5%)
Short diameter (mm)	4.0 (2.0–9.0)	5.6 (2.5–11.8)
Distance to the rectal wall (mm)	9.0 (1.5–27.7)	5.9 (1.7–12.1)
Distance to the mesorectal fascia (mm)	4.3 (0.5–14.5)	5.7 (0.3–10.0)
Distribution
Below the distal tumor	5	0
Level of the tumor	129	17
≤ 5 cm proximal to the tumor	144	16
> 5 cm proximal to the tumor	2	0

PS primary surgery, NT neoadjuvant treatment

Discussion

Imaging has proven to be an inaccurate modality for detecting metastatic LNs in rectal cancer. Most previous studies contained pathological nodal staging from the number of positive LNs found in the entire mesorectum, lacking the pathological ground truth of each LN detected in the images. LNs status can only be indirectly predicted by primary tumor features, hindering further analysis of LN features.

Thus, a deeper level of analysis is required at the LN level, matching individual LNs seen on images to their precise histological counterparts, to treat each LN as a labeled target. However, further precise matching is confounded by multiple factors during imaging evaluation and histological verification, including: (1) Accurate localization of individual LN in the complex structure of the mesorectal anatomical package; (2) Deformation of surgical resection tissue relative to its in vivo conformation and formalin fixation-induced tissue shrinkage; (3) Difference between slice thickness and MRI slice thickness; (4) Macroscopic visual registration of tissue specimen slices and MRI image slices. Although several related studies based on node-by-node comparisons have been published so far, there are significant differences in schemes, protocols and parameters. The reported matching accuracy also varies, making an accurate histological-imaging correlation technically challenging (Zhang et al. 2017; Brown et al. 2003; Koh et al. 2004, 2005, 2010; Park et al. 2014; Rutegård et al. 2020; Lahaye et al. 2008; Lambregts et al. 2011, 2013; Heijnen et al. 2014) (Table 3). Through this report, we propose a technique to achieve LN histology-imaging correlation and comprehensively research benign and malignant LNs based on node-by-node discovery. Using this technique as part of a prospective study makes it possible to analyze LN features at the LN level, ultimately improving the accuracy of rectal N-staging and treatment efficacy in the future.

Table 3.

Previous studies on node-by-node comparison of mesorectal lymph nodes imaging findings to histopathology

Study	Study object	Matching method	LN	Match success rate
Brown et al. (2003)	MRI	Specimen MRI	437	NA
Koh et al. (2004)	(USPIO)-enhanced MRI	Specimen MRI	74	94%
Koh et al. (2005)	MRI	Specimen MRI	85	90%
Lahaye et al. (2008)	(USPIO)-enhanced MRI	Anatomic map	236	71%
Koh et al. (2010)	(USPIO)-enhanced MRI	Specimen MRI	126	94%
Lambregts et al. (2011)	Gadofosveset-enhanced MRI	Anatomic map	521	NA
Lambregts et al. (2013)	Gadofosveset-enhanced MRI	Anatomic map	387	75%
Park et al. (2014)	MRI	Ex vivo ultrasono-guided	205	91.10%
Heijnen et al. (2014)	Gadofosveset-enhanced MRI	Anatomic map	543	NA
Zhang et al. (2017)	MRI	Specimen MRI	290	64.10%
Rutegård et al. (2020)	FDG-PET/MRI	Specimen MRI	92	47%

LN lymph node, NA not available, MRI magnetic resonance imaging, USPIO ultrasmall superparamagnetic iron oxide

Preoperative imaging assessment using the proper modalities is the most important. Most previous researchers have used MRI as a tool to detect LN metastasis. T2WI magnetic resonance imaging has been proven to accurately display factors such as the mesorectal fascia and the depth of tumor invasion (Beets-Tan et al. 2018). However, the standard T2WI typically has few layers and low LNs contrast, thus may confuse tiny LNs with vascular deposits (Kim et al. 2004). In contrast, DWI derives its contrast from differences in the cellular density between tissues, which can depict LNs as hyperintense, making it a valuable LNs detection technique (Kwee et al. 2008). Research has shown that DWI is superior to T2WI in detecting LNs, and when combined with T2WI, DWI exhibited excellent sensitivity in identifying small LNs (Heijnen et al. 2013; Mizukami et al. 2011). Therefore, by combining DWI and T2WI, it is possible to accurately locate individual LN in the complex structure of the mesorectal, avoiding misjudgment or omission.

The processing of specimens is crucial. The operation method should follow the standardized evaluation process for CRM (Dworak 1991). One of the key points is to mark the midline of the mesorectal long axis as an anatomical landmark for the isolated tissue, because it will be challenging to discern its exact position after complete fixation; failure to ensure that slicing is always perpendicular to the mesorectal long axis will eventually cause loss of anatomical consistency with the image. To ensure the physiological morphology of the mesorectum, we pinned the rectum's proximal, middle, and distal ends on the foam board and fully straightened them under tension to avoid bending or shrinkage of the mesorectum, ensuring that the mesorectal fat remains tissues are evenly distributed during sectioning. The tissue specimens will then be sliced into 3 mm thick slices. This step is usually done manually with a knife and thus the quality of the sections may be affected by the angle, speed, etc. of sectioning. Slices thickness is also often not entirely precisely controllable. However, it should be noted that this is different from matching MRI images to histological microscope images (Alyami et al. 2022). The latter requires thin sections with a microtome after dehydration and embedding in wax, and finally correlates with the MRI images, thus requiring extremely high precision and section thickness control. In fact, there is little risk of misalignment of slice and MR images in a visual registration environment based on the macroscopic tissue section. Ideally, cutting with a rotary microtome may yield more regular sections, but poor ex vivo tissue stability during cutting may lead to section distortion and is thus not recommended.

Another improvement of this study is the registration method, the way to improve the precision and accuracy of image registration, including visual assessment, landmark selection, etc. (Maintz and Viergever 1998). In the validation based on histological and MRI images, accuracy can be improved by modifying standard clinicopathological protocols. Brown et al. (2003) first began to use the supplementary imaging methods. They matched tissue slices with MRI and ex vivo MRI images. The study obtained satisfactory matching accuracy. The advantage of this method is that the ex vivo images are not affected by artifacts of physiological activity, while having greater spatial resolution and higher signal-to-noise ratio than in vivo images. However, in practice, since the two MRI examinations involve massive imaging images that need to be compared with the corresponding histological, thus it is hard to conduct a large sample study. We referred to the strategy used by Lahaye et al. (2008), which entailed using standardized anatomical mapping as a template for matching. This, however, is not suitable for all situations; for example, one would not perform this in case of crowded mesorectal LNs, since no quantitative measurement is produced, a purely visual assessment may risk a mismatch. We have improved upon this by suggesting vertical and horizontal anatomical maps. The basic principle of the vertical map is to judge the cranial-caudal distribution of the mesorectal nodes in relation to the primary tumor, and determine the corresponding histological slice level of each LN and the total number of LNs at the corresponding slice level. When matching vertically, first marking the inferior tumor margin as the starting point (point 0), then numbering sequentially from the starting point to proximal, the actual distance corresponding to adjacent numbers is 3 mm. The radiologist used the sagittal and axial images to estimate the distance of each node from the inferior tumor margin and marked all the LNs at the corresponding position by distance. Since the actual spacing between numbers is consistent with the final slice thickness (3 mm), pathologists will easily find the target LN on the corresponding slice. The horizontal map is a landmark-based validation method. When matching horizontally, we outlined the size and shape of each node, the minimum nodal distance to the mesorectal fascia, rectal wall, and possible positional relationship with the surrounding blood vessels and adjacent LNs. By filtering irrelevant and redundant image information, a simplified map of cross-section imaging was obtained (Fig. 2). Since there are no anatomical features that can be visually traced across modalities within the mesorectum, the horizontal map approximation establishes fiducial markers to quantify the registration accuracy of tissue sections and MR images. Compared with the previous methods, we have modified the method for the anatomical map, greatly improved practicability while simplifying the process of repeatedly viewing images, and eliminating additional imaging checks on specimens.

This study has some limitations. First, the case scale is small, and a more extensive prospective study is underway to ascertain the accuracy of the technique in detecting nodal metastases. Second, due to spatial resolution limitation, precisely located LNs ≤ 2 mm is difficult when analyzed node-by-node. Heijnen et al. addressed the spatial resolution limits by using dedicated nodal imaging MR sequences with 1 mm isotropic voxels (Heijnen et al. 2016). Indeed, compared to finding tiny LNs on images, the greater challenge lies in how to match from pathological specimens; tiny LNs are susceptible to different times and spaces in the mesorectal, so we did not enroll LNs ≤ 2 mm on imaging. This also explains the higher matching success rate of 90.5% (313/346), but a large discrepancy between the number of LNs identified by histopathology 353/660 (54%). Among the unmatched LNs, the large number located at the root of the mesorectum (i.e., nodes high up along the superior rectal artery and vein); because they lie in a loose mobile mesentery, some larger LNs can be matched by size and spatial location, but unable to achieve sufficient accuracy, so we exclude them. Nevertheless, due attention should be paid to this area because of its high LNs yield (Langman et al. 2015). Moreover, we excluded patients with > 15 perirectal LNs on preoperative MRI because it entails the risk of mismatching in cases of crowded LNs within the mesorectum.

Finally, this study aims to provide a standardized and easily generalizable technique, with the anatomical map being accessible and low economical cost. Although we have devised a range of methods to accommodate routine clinical tissue processing and improvements over traditional matching methods. However, there are still many factors affecting reliable matching during imaging evaluation, tissue section processing, and final matching, and these limitations need to be improved and optimized in the future.

Conclusion

The technique to match mesorectal LN imaging findings to histopathology was feasible and effective. It simplified the technical method and provided an excellent success rate. This technique provides a standardized approach to follow-up studies at the lymph node level. Validation of its efficacy requires additional investigation in larger cohorts.

Abbreviations

MRI

Magnetic resonance imaging

DWI

Diffusion weighted imaging

LNs

Lymph nodes

CRM

Circumferential resection margin

MRF

Mesorectal fascia

Author contributions

ZZ designed the project, developed the search strategy and wrote the manuscript. XM, YZ, MW, XD and XY participates in nodal matching and radiologic-pathologic comparison. ZW design the conception, revise critical intellectual content. All authors read and approve the final manuscript and agree to be accountable for all aspects of work to ensure that questions regarding accuracy and integrity investigated and resolved.

Funding

This study was supported by Department of Science and Technology of Sichuan Province (Award Number 2021YFS0025); 1·3·5 project for disciplines of excellence, West China Hospital, Sichuan University (No. 2019HXFH031; 20HXJS003); Post-Doctor Research Project, West China Hospital, Sichuan University (2021HXBH033); the Ethicon Excellent in Surgery Grant (EESG) (No. HZB-20190528-4).

Data availability

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Conflict of interest

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

Ethics approval

This study was approved by the Medical Ethics Committee of West China Hospital.

Consent to participate

Informed consent was obtained from all individual participants included in the study.