Sage Journals: Discover world-class research

Abstract

Through meticulous examination of lymph nodes, the stage and severity of cancer can be determined. This information is invaluable for doctors to select the most appropriate treatment plan and predict patient prognosis; however, any oversight in the examination of lymph nodes may lead to cancer metastasis and poor prognosis. In this review, we summarize a significant number of articles supported by statistical data and clinical experience, proposing a standardized evaluation protocol for lymph nodes. This protocol begins with preoperative imaging to assess the presence of lymph node metastasis. Radiomics has replaced the single-modality approach, and deep learning models have been constructed to assist in image analysis with superior performance to that of the human eye. The focus of this review lies in intraoperative lymphadenectomy. Multiple international authorities have recommended specific numbers for lymphadenectomy in various cancers, providing surgeons with clear guidelines. These numbers are calculated by applying various statistical methods and real-world data. In the third chapter, we mention the growing concern about immune impairment caused by lymph node dissection, as the lack of CD8 memory T cells may have a negative impact on postoperative immunotherapy. Both excessive and less lymph node dissection have led to conflicting findings on postoperative immunotherapy. In conclusion, we propose a protocol that can be referenced by surgeons. With the systematic management of lymph nodes, we can control tumor progression with the greatest possible likelihood, optimize the preoperative examination process, reduce intraoperative risks, and improve postoperative quality of life.

Keywords

lymph nodes medical imaging lymphadenectomy immunotherapy machine learning statistical models

Background

Standards for Examined Lymph Node Counts

The examined lymph node (ELN), a prognostic factor for solid tumors, such as colorectal cancer, lung cancer, and gastric cancer, represents whether or not lymph node dissection at the surgical site was adequately performed. This factor is important for cancers characterized by more lymph node metastasis, more recurrence and poorer five-year overall survival and is considered a key indicator for evaluating surgical quality.¹ The guidelines from the National Comprehensive Cancer Network (NCCN) recommend that a minimum number of examined lymph nodes should be established for each tumor. It allows standardization and proceduralization, ensures surgical quality control, and reduces operative time and postoperative complications. For example, as early as 1990, the World Congress of Gastroenterology in Sydney recommended at least 12 ELNs for colorectal cancer. This standard was not only endorsed by the College of American Pathologists (CAP) and the American Joint Committee on Cancer (AJCC) but also incorporated into NCCN evidence-based guidelines.^2-4 Recently, there has been another opinion about escalating this standard to 15 lymph nodes.⁵ For gastric cancer, the Japanese Classification of Gastric Cancer from the Japanese Gastric Cancer Association (JGCA) and the eighth edition of the AJCC Manual of Gastric Cancer both recommend that the number of ELNs be at least 16.^6,7 Some researchers have also questioned the appropriateness of this standard. Further investigations revealed that 20 nodes are needed, while others reported that 25 nodes are required.^8,9 For non-small-cell lung cancer (NSCLC), the International Association for the Study of Lung Cancer (IASLC) suggested a minimum of 6 ELNs for precise nodal staging, which included 3 at the N1 station and 3 at the N2 station.¹⁰ The European Society for Thoracic Surgery (ESTS) suggested that at least 6 hilar or mediastinal lymph nodes be dissected.¹¹ A study published in the Journal of Clinical Oncology stated that the optimal number of nodes used to stratify node-negative and node-positive NSCLC should be 16.¹² However, a subsequent study argued for lowering the standard to 10.¹³ For head and neck squamous cell carcinoma, the optimal number ranges from 16 to 18 to 19 to 20. A conclusion has not yet been reached.^14-18 For renal cell carcinoma, the threshold can be as low as 3.¹⁹ The minimum number of ELNs for pancreatic cancer has not been set, as there is still a great deal of variation between guidelines. First, it was not addressed in the NCCN guidelines. Second, according to the AJCC and the Union for International Cancer Control (UICC), a minimum of 10 lymph nodes should be dissected for staging. Third, according to the European Society for Medical Oncology (ESMO), 15 or more is favored.²⁰ A study published in the Annals of Surgery proposed a minimum number of 11 and an optimal number of 19 as new validated standards for intraoperative node dissection and postoperative patient stratification in pancreatic cancer patients.²¹ For esophageal squamous cell carcinoma, the minimum number and the optimal number were 14 and 18, respectively.²² (Table 1).

Table 1.

Standards for ELN Count in Various Cancers.

Cancer	Threshold	Source
Colorectal cancer	12	CAP; AJCC; NCCN
Colorectal cancer	15	Guan⁵
Gastric cancer	16	JGCA; AJCC
Gastric cancer	20 or 25	Okajima⁸, Sura⁹
NSCLC	6	IASLC; ESTS
NSCLC	16 or 10	Liang¹², Zhu¹³
HNSCC	16, 18, 19, 20 or more	Divi¹⁴, Ebrahimi¹⁵, Kuo¹⁶, Lee¹⁷, Agrama¹⁸
Renal cell carcinoma	3	Zhuang¹⁹
Pancreatic cancer	10	AJCC; UICC
	15 or more	ESMO
	Minimum: 11 Optimum: 19	Huang²¹
ESCC	Minimum: 14 Optimum: 18	Tian²²

Proper management of ELNs can greatly improve the prognosis of cancer patients. In contrast, it can have consequences, including occult node metastasis, lymph node micrometastasis, stage migration and immune impairment.

Occult Node Metastasis and Lymph Node Micrometastasis

Occult node metastasis refers to positive lymph node (PLN) missed during preoperative palpation and imaging but accidentally found in postoperative tissue sections, which commonly occurs in patients at clinical stage N0/N1.^23-25 Similarly, lymph node micrometastases are clusters of microscopic tumor cells sneaking in lymph nodes; these clusters are too small to be detected on preoperative examination and are even overlooked during routine macroscopic pathology. These clusters of cells are defined by the AJCC as being between 0.2 mm and 2.0 mm in diameter. Further examination requires more sensitive techniques, such as immunohistochemical staining, flow cytometry, and reverse transcriptase polymerase chain reaction.⁷ Unfortunately, even with the application of the above technologies, a high detection rate cannot be achieved. Research indicates that among patients tested for lymph node-negative status through carcinoembryonic antigen-specific nested RT‒PCR, 54% end up with micrometastatic disease at surgery.²⁶ It is vital to detect lymph node micrometastasis as much as possible to make accurate clinical decisions and to provide prompt intervention in early-stage patients with this risk factor.

Stage Migration

Occult nodal metastasis and lymph node micrometastasis are characteristic of stage migration, with an insufficient number of ELNs being the cause of stage migration. Stage migration is a type of Will Rogers phenomenon, which in this context refers to changes in the N stage. More specifically, a group of node-positive patients may be misclassified as node-negative in clinical practice due to inadequate preparation or operation, representing the occurrence of missed diagnoses.^27,28 With further treatment, micrometastasis is identified, and patients previously categorized as “good stage” are assigned to “bad stage”. Although the prognosis of reclassified patients was worse than that of the remaining patients of “good stage”, it was better than that of other patients of “bad stage”. Survival outcomes appeared to improve in each group, while the individual outcomes remained the same.

The medical oncologist determines whether postoperative radiotherapy or chemotherapy is necessary for a patient according to his or her N stage. Those prone to stage migration will neither receive an appropriate treatment plan from the oncologist nor provide credible treatment feedback with the oncologist. Statistically, the reliability of survival data for the pN0 population is compromised. In short, stage migration tampers with the whole picture.

Immune Impairment

Immune impairment refers to the weakening of anti-tumor immunity due to the expanded dissection of tumor-draining lymph nodes (TdLNs). This finding is contrary to the generally accepted theory that “the more lymphadenectomy, the better.” Since these nodes undergo primary tumor antigen exposure and subsequent antigen-specific immune activation, the roles of harboring effector lymphocytes cannot be exaggerated.²⁹ In addition to resulting in increased susceptibility to postoperative infections, immune impairment further hampers the implementation of adjuvant immunotherapy due to the lack of appropriate treatments.

These three consequences cover three phases: preoperative examination, intraoperative lymphadenectomy, and postoperative therapy for cancer patients. The questions to be answered include how to detect as many PLNs as possible by noninvasive approaches preoperatively, how to remove as many PLNs as possible and how to preserve as many immune-dependent lymph nodes as possible intraoperatively. These three questions have received close attention from oncologists in recent years. This review summarizes multi-institutional, multinational, retrospective and prospective cohort studies over the past two decades, covering the entire research process from patient selection to data analysis to outcome assessment. It is distilled into a comprehensive evaluation protocol for cancer patients, aimed at providing guidance for clinical research and practice (Figure 1).

Figure 1.

The evaluation protocol for examined lymph nodes in cancer patients. The whole protocol involved preoperative examination, intraoperative lymphadenectomy and postoperative immunotherapy. Preoperative examination aided by deep learning models categorizes patients into three groups: clinically node-negative, low-risk and high-risk in clinically node-positive. Those at risk will receive the necessary neo-adjuvant therapy. Following the preliminary screening, patients will undergo further surgical procedures. Lymphadenectomy guided by statistical models categorizes patients into three groups: pathologically node-negative, low-risk and high-risk in pathologically node-positive. Those at risk will receive the necessary postoperative immunotherapy.

Preoperative Examination

Ultrasound, positron emission tomography (PET), computed tomography (CT) and magnetic resonance imaging (MRI) are commonly used as noninvasive preoperative approaches for screening lymph nodes in various cancers. The combination of PET/CT has been recognized as the most reliable diagnostic solution for the diagnosis of occult lymph nodes in lung cancer and gastric cancer, with higher sensitivity and specificity in staging than CT alone and higher sensitivity than PET alone.^30-32 This is because CT provides anatomical information to determine the location of the node, and PET provides metabolic information to determine whether the node is positive. The two complement each other to form a more complete picture. In head and neck squamous cell carcinoma, a meta-analysis showed no difference in the sensitivity or specificity of the above imaging approaches, except that CT displayed greater specificity than ultrasound.³³ However, in addition to intratumor heterogeneity, there are inherent limitations in human vision. Subjective differences only make it more difficult to achieve reproducibility and consistency in examination.³⁴ Relying solely on manual inspection will result in a large number of missed cases of occult node metastasis.³⁵ In NSCLC, 12.9%-39.3% of lymph node metastases are not identified by PET/CT.^36-39 In gastric cancer, diffusely growing and mucus-containing tumor types are likely to be falsely identified as negative for low uptake of fluorodeoxyglucose.⁴⁰ In head and neck squamous carcinoma, either PET/CT or MRI lacks accuracy in the diagnosis of occult lymph nodes.⁴¹ With the development of molecular imaging toward quantitative imaging, radiomics under the background of deep learning has integrated multiple disciplines, such as medical imaging, clinical medicine, computer science, and statistics. Convolutional neural networks are able to extract high-throughput features from images in an automated and repeatable manner, converting them into high-dimensional data; however, these features are unobtainable under human vision. With respect to the detection of occult lymph nodes, the performance of radiomics has been proven to be superior.^42-44

For model establishment and model validation, an internal cohort and an external cohort are necessary for the patient population, with the internal cohort divided into a training set at 80% and a validation set at 20%. The additional prospective cohort allows observation of whether the model is feasible for clinical use and enables further validation. First, the input data included the region of interest of the original image and various parameters, such as tumor size, center location, merging rate and metabolic value. Subsequently, a deep learning model is used for spatial feature extraction, including texture features, first-order statistical features, and higher-order statistical features. Popular neural network structures for image recognition include ResNet, ImageNet, U-Net, and DenseNet. The multimodal algorithms used integrated image information, clinical information, and follow-up records for quantitative analysis. It is important to note that overfitting and poor generalization should be avoided when using multimodal data.

Current PET/CT radiomics research on occult lymph nodes is focused on lung cancer. In the study by Wallis et al, a model based on the U-net not only identified PLNs at the expert level but also localized the thoracic region and suspicious regions in the complete image without previously outlining the ROI by a radiologist.⁴⁵ A review on PET and CT in the detection of occult lymph nodes in lung cancer patients involved two studies that included the combination of PET/CT data, one based on deep learning and the other based on semi-automated machine learning that required manual extraction of image features. Both demonstrated that compared to using either one kind of imaging data or clinical-histopathological data alone, their models have the best potential to predict lymph node micrometastasis.^46-48

A recent study selected PET/CT images from a cN0 population who had undergone curative surgery for primary NSCLC as data resources, and the deep learning architecture used was finalized after comparison with other architectures, such as ResNet18.⁴⁹ The model is named Deep Learning Nodal Metastasis Signature. The largest Area Under the Precision-Recall Curve was achieved not only in comparison with PET images alone and CT images alone but also with the prediction model-based clinical variables and the scoring systems of senior and junior radiologists, suggesting that the ability of the model to differentiate between positive and negative cases was excellent.⁵⁰ The calibration curves and the decision curve showed that the model was quite close to the real situation and was clinically useful. In addition to effectively stratifying early-stage patients, it provides reliable clinical opinions on whether sublobectomy or lobectomy, elective nodal dissection or systematic nodal dissection (SND), or postoperative adjuvant therapy or a lack of postoperative adjuvant therapy should be adopted for low-risk and high-risk patients.

Apart from PET/CT, deep learning models have demonstrated superiority in predicting lymph node micrometastasis in MRI and ultrasound examinations for cervical cancer and breast cancer.^51-54

In the detection of cervical lymph nodes, a ResNet50-based deep learning model was robustly able to distinguish between lymph node metastasis and lymphoma using PET/CT images in a patient population with a chief complaint of enlarged lymph nodes.⁵⁵ Although this is not a study targeting lymph node micrometastasis, the developed model can be one of the preferred options for a variety of cancers that initially present with enlarged superficial lymph nodes, such as head and neck squamous carcinoma, thyroid cancer, and breast cancer. Subsequently, other deep learning models for the detection of occult lymph nodes can be used to refine the overall lymph node examination, thus maximizing the use of whole images for preoperative examination. Invasive examinations such as ultrasound-guided fine needle aspiration cytology have advantages in the examination of superficial lymph nodes, but drawbacks are also evident. First, they are not applicable for lymph nodes smaller than 7-8 mm in diameter, ie, they have limitations in the diagnosis of lymph node micrometastasis. Second, informed consent from the patient for invasive procedures is needed. Third, the procedure is highly dependent on the physician's skill. Fourth, it is likely to irritate the tumor site.^41,56-59

Intraoperative Lymphadenectomy

Multivariate Regression Analysis of the ELN Count

At the beginning of this century, multiple retrospective studies confirmed a close relationship among ELN count, survival and lymph node metastasis by multivariate regression analysis. Patient data were obtained from a single institution, the National Cancer Database (NCDB), and the Surveillance, Epidemiology, and End Results (SEER) database. The total size of the SEER database is smaller than that of the NCDB database, but it is open to the public, easily accessible, and widely used in domestic studies. All patients were admitted early for primary surgery. The demographic characteristics included sex, age, race and ethnicity; clinical characteristics, such as comorbidities, primary tumor site, and tumor size; and pathological characteristics, such as Tumor Node Metastasis (TNM) classification, number of ELNs, number of PLNs, and involvement of margins. Inclusion criteria usually consisted of age greater than eighteen years, tumor confined to the primary site, and resection of at least one or more lymph nodes; exclusion criteria usually consisted of failure to undergo surgery, perioperative death, presence of distant metastasis, history of neoadjuvant therapy or history of cancer, and failure to follow-up; and evaluation criteria usually consisted of disease-free survival, disease-specific survival and overall survival. To prevent stage migration, some researchers excluded patients with few ELNs when performing multiple sensitivity analyses to control for statistical bias.⁹

Researchers have reached a consensus that the greater the ELN count is in early-stage patients with pathologically negative nodes, the lower the recurrence rate and the higher the survival rate, as well as the greater the likelihood of detecting occult lymph nodes. This suggests that incomplete lymphadenectomy, inaccurate pathological evaluation, and incorrect histopathological staging have already lurked in the original patient population, leading to missed diagnoses of patients with PLNs and failure to receive postoperative adjuvant therapy; on the other hand, the improvement in survival due to the increase in ELN count reflects an increase in correct staging and a decrease in occult lymph nodes.^18,60-64 In patients with pathologically positive nodes, researchers agree that the greater the PLN count or the greater the ratio of PLN count to ELN count is, the lower the survival rate. After controlling for PLN count, the survival rate still increased with ELN count.^65-67 Thus, the ELN count is considered an important prognostic variable. The greater the number of nodes dissected is, the greater the likelihood of accurate staging, and the better the prognosis.

In addition to lymph node count being positively associated with survival, studies on colorectal cancer have shown that lymph node size (maximum long-axis diameter) is also positively associated with survival, with one study identifying 10 mm as the threshold based on the difference in five-year overall survival, which may be due to the stronger anti-tumor immune response of larger lymph nodes.^68,69 Further research is required to confirm the correlation between the size of the lymph nodes and the occurrence of lymph node metastasis. Emerging single-cell analysis may provide novel insight into the relationship between the two from aspects such as transcriptomics and proteomics and establish optimal thresholds that can distinguish the size of lymph nodes according to the most typical difference.

Thresholds of the ELN Count

Despite the clarification of the relationship between ELN count and prognosis, guidance for practice is needed. In addition to the exhaustive method of selecting thresholds by multivariate regression analysis mentioned above, researchers have used larger population datasets based on more statistically convincing methods to explore thresholds of ELN count for each type of solid tumor. They aim to distinguish between clinically node-negative and clinically node-positive patients and to eliminate stage migration effects with the greatest likelihood of success. Larger population datasets are often obtained from multiple databases and multiple low- to high-volume medical centers. Typical examples of the various statistical methods used to calculate thresholds are listed below (Table 2).

Table 2.

Characteristics of the Models for Evaluating the ELN Count Thresholds.

Model	Cancer	Tool	Key measurement	Function	Source
Hypergeometric Distribution-Based Model	Breast cancer, colorectal cancer		The number of positive nodes; the total number of nodes; the maximum long-axis diameter of nodes	For all patients: evaluating the minimum ELN count and the count of residual PLNs	Kiricuta⁷⁰;Iyer⁷¹;Suzuma⁷²;Joseph⁷³
Hypergeometric Distribution-Based Model	Breast cancer, colorectal cancer			For node-positive patients: evaluating the minimum count and the maximum long-axis diameter of ELN for patient stratification	Kiricuta⁷⁰;Iyer⁷¹;Suzuma⁷²;Joseph⁷³
Beta-binomial Distribution-Based Model	Colorectal cancer, signet ring cell carcinoma, gallbladder carcinoma, esophageal squamous cell carcinoma, papillary thyroid carcinoma	VGAM package from R	The total number of nodes for a node-positive patient	For all patients: evaluating the minimum ELN count	Gönen⁷⁴;Yu⁷⁵;Shariat⁷⁶;Zheng⁷⁷;Zhang⁷⁸;Robinson⁷⁹
Survival Data-Based Model	NSCLC	LOWESS function from R;Chow test from SAS;X-tile	The number of positive nodes;the total number of nodes;odds ratios (ORs; stage migration) and hazard ratios (HRs; overall survival)	For all patients: evaluating the ELN count as the breakpoint of survival	Liang¹²;Zhu¹³

Hypergeometric Distribution-Based Model

This model was pioneered in 1992 in a study by Kiricuta et al⁷⁰ Based on hypergeometric distribution and Bayes’ theorem, the number of residual PLNs after axillary lymph node dissection (ALND) can be estimated with 90% certainty. A total of 1446 T1 (n = 831), T2 (n = 565), and T3 (n = 50) breast cancer patients who underwent ALND were included. After categorizing the calculation according to T stage, the model yields, on the one hand, the minimum number of lymph nodes that need to be sent to avoid the presence of residual PLNs for each T stage with 90% certainty, and on the other hand, the number of PLNs that may be left after a certain number of lymph nodes have been sent and a certain number of PLNs have been identified for each T stage with 90% certainty. The mathematical model provides both preoperative instructions and postoperative evaluation of lymphadenectomy. It is able to identify pN0 patients with 90% certainty and to supplement adjuvant therapy for patients with incompletely removed PLNs.

In 2000, Iyer et al used this model for the evaluation of patients with detected PLNs.⁷¹ Patients with primary breast cancer are usually categorized into three levels according to PLN count: pathologically negative, 1-3 PLNs, and greater than or equal to 4 PLNs, with the third level being an indication for postoperative radiotherapy.^67,80 According to their study, the minimum ELN counts required to achieve 90% accuracy were 19, 20, and 20 for T1 patients who were predicted to have 1-3 PLNs, respectively; the minimum ELN counts were all 20 for T2 patients who were predicted to have 1-3 PLNs. For T1 and T2 patients in whom 1-3 PLNs were already found, the study also recommended the minimum ELN count to ensure that no more (greater than or equal to 4) PLNs remained. This study aimed to accurately stratify breast cancer patients and to ensure the necessity of supraclavicular regional radiotherapy after mastectomy. Moreover, as mentioned in section 2.1, lymph node size is associated with survival rates. In 2001, Suzuma et al noted that Kiricuta et al neglected the impact of lymph node size.⁷² The mathematical model established in their research was inconsistent with the previous model. In addition to the ELN count and PLN count, they also incorporated the parameter of the maximum long-axis diameter of the lymph nodes. When data were collected from patients after ALND, the size of each node was recorded under a microscope. Since the study included T1 and T2 breast cancer patients with lymph node metastasis, the results of the study illustrate the minimum ELN count and the corresponding maximum long-axis diameter of the lymph node required to detect a PLN with 90% certainty rather than to avoid the presence of residual PLNs. The results complement those of the study by Iyer et al Together, these three studies complete the guidance for ALND in early-stage breast cancer patients.

In 2003, the model of Kiricuta et al was also applied to estimate the number of residual positive mesenteric lymph nodes in patients with colorectal cancer.⁷³ Compared with the previous study, this study included patients at intermediate and terminal stages, basically T3 and T4; it only accomplished the first half of Kiricuta et al's work—a minimum of 40 ELNs to achieve an 85% probability of pathological negative results in T3 patients and a minimum of 30 ELNs in T4 patients. The status of patients with residual PLNs was not determined, probably because postoperative adjuvant therapy is routinely implemented for colorectal cancer patients at intermediate and terminal stages.

Beta-Binomial Distribution-Based Model

This model is based on the maximum likelihood estimation, which is used to find the phylogenetic tree capable of generating observed data with a higher probability, ie, it is used to determine the ELN count, by which surgeons perform lymphadenectomy to have a greater probability of avoiding missed diagnoses of PLNs.⁸¹ As early as 1973, Griffiths first used the beta-binomial distribution as the basis for modeling the incidence of non-communicable diseases.⁸² In 2009, maximum likelihood estimation was applied to the prediction model of lymph node metastasis in colorectal cancer, and the Nodal Staging Score (NSS) was formulated on the premise that nodal staging follows a beta-binomial distribution, marking the beginning of this type of research.⁷⁴ In addition to a series of validation studies, studies of signet ring cell carcinoma, gallbladder carcinoma, esophageal squamous cell carcinoma, and papillary thyroid carcinoma were carried out, all of which confirmed the prognostic value of the model in pN0 patients.^75-78

The key in the NSS is to estimate the probability of false negatives (P(FN)). This step is performed in the population of patients with positive nodes, so the data cleansing needs to start by filtering out the population of positive nodes from the total population and excluding the population of patients who may undergo lymph node biopsy rather than lymph node dissection; in other words, the ELN count is required to be more than one. According to the beta-binomial distribution model, P(FN) and the ELN count follow the formula:

[P (F N) = \frac{B (α, β + m)}{B (α, β)}]

where B(.) represents the beta function, α and β are the shape parameters of the probability density function and m is the ELN count. The VGAM package in R has a built-in maximum likelihood estimation program, so the α and β parameters can first be automatically fitted first to the ELN count in the population of positive nodes. Subsequently, the ELN count and the parameters are included in the formula to produce a percentage distribution curve for the probability of false-negative lymphadenectomy in all patients with at least one positive node. It is important to specify that the establishment of this probability density function relies on three assumptions: 1. There are no false positives; 2. there is no selective bias, and all ELNs are selected with equal probability; and 3. the algorithm has the same sensitivity to both true positives and false negatives.⁷⁴

After deriving the probability of false negatives corresponding to each number of ELNs, the number of false negatives in each group can be derived to calculate the corrected prevalence of true positives and to estimate the true prevalence of nodal disease in the entire study population. Finally, the number of true positives was subtracted from the total number of patients to calculate the node staging score of true negatives as a function of the ELN count. Furthermore, to achieve accurate staging, the total population was divided into different subgroups according to the T1, T2, T3, and T4 stages during data cleansing, and then applied to the NSS. According to the different T stages, the minimum number of ELNs was determined for the different populations. Following this number can achieve no residual PLNs with 90% certainty.⁷⁹

In the section discussing the quantitative relationship between ELNs and residual PLNs, the establishment of a beta-binomial distribution-based model and the derivation of the shape parameters are robust and reasonable, while Assumption 2, which the model relies on, is controversial in practice compared to the other two assumptions. This is because in most public databases, such as the SEER database or the NCDB, information on the ELN does not include information on lymph node sites. During END, surgeons usually have different sampling preferences for lymph nodes at different sites, eg, lymph nodes that are closer to the tumor site are more likely to be dissected than those further from the tumor site. A study by Lee et al also suggested that different approaches to the management of lymph nodes may result in different lymph node yields, so the heterogeneity of intraoperative dissection and microscopic examination should be taken into account.¹⁷ However, Gil et al reported that there was no significant correlation between the ELN count and the PLN count (correlation coefficient R = 0.3), which eliminates the suspicion that the status of the lymph nodes is affected by selection bias.⁸³ Although the lymph node site has no effect on the PLN count, it does have an impact on patient prognosis. One study showed that the status of lymph nodes in the right colon was more closely related to the risk of death than that of lymph nodes in the left colon.⁸⁴ This highlights the need for separate thresholds for lymph node subsites.

Survival Data-Based Model

Both of the above mathematical models estimate a threshold based on the distribution of the ELN count, which is subsequently verified with survival data. Other researchers have analyzed the ELN count and survival data in the first step. In the second step, locally weighted regression or X-tile statistical analysis software was used to perform structural break point analysis or cutoff value analysis of the effect of the ELN count on overall survival without any assumptions, and in the third step, the calculated threshold was used to verify the effect on survival in a secondary way.^{12,13,19,21,22}

Locally weighted regression (LOWESS) is a non-parametric regression method used in statistics and machine learning that captures complex nonlinear relationships in a dataset. It achieves a localized fit by assigning higher weights to points near the prediction point and lower weights to points far from the prediction point. The presence of structural break points in a dataset is subsequently determined by comparing the regression coefficients of the fitted curves on both sides of each potential structural breakpoint by the Chow test for significance. In 2017, the method was first applied by Liang et al for NSCLC.¹² The patient populations included were early-stage, multi-institutional, foreign and domestic NSCLC patients. There were 38 806 patients from the SEER database and 5705 patients from Chinese institutions. The structural break point from the training set passed independent internal and external validation. It is important to note that this study can only demonstrate that dissecting lymph nodes by the ELN count in the vicinity of the structural break point improves patient mortality but does not support that the ELN count exceeds the structural break point too much. In the curve to the left of the structural breakpoint, an increase in the ELN count avoids erroneous staging and reduces residual PLNs; in the curve to the right of the structural breakpoint, an increase in the ELN count is associated with a poorer prognosis.

X-tile is a graphical statistical analysis software designed to help identify which cutoff values separate patients into clusters with different survival outcomes by analyzing biomarker expression levels and survival data from a large number of patient samples.⁸⁵ Zhu et al considered the standard set by Liang et al to be questionable and included more early-stage NSCLC patients with the same inclusion and exclusion criteria—52 099 from the SEER database and 3002 from Chinese institutions.¹³ They employed X-tile software to exhaustively search for various numbers of ELNs and eventually identified the number of ELNs that showed the most significant difference (the largest chi-square value, relative risk, and smallest P value) as the cutoff value. This cutoff value was confirmed to be an independent prognostic factor after excluding confounders and homogenizing the baseline by Propensity Score Matching. Overall survival in early-stage NSCLC patients also decreased in accordance with the results of Liang et al's research. They considered this cutoff value to be up-to-date and attributed the difference to the database version and the TNM classification version; however, they did not mention a possible impact of the methodological difference between the two studies.

Lymph Node Ratio or Lymph Node Density

The lymph node ratio (LNR) or lymph node density (LND) (PLN count/ELN count) is becoming a viable prognostic option for node-positive patients.^83,86-94 The superiority of this fraction lies in its ability to simultaneously evaluate three factors: 1. tumor factors (PLN count); 2. treatment factors (ELN count); and 3. staging factors (completeness of the sampling procedure, including those related to the surgeon and pathologist).⁸³ Like for ELNs, the corresponding threshold for each cancer is also available through multivariate regression analysis and X-tile software. The LNR has also been proven to be superior to the traditional TNM classification for predicting disease-free survival, disease-specific survival and overall survival. For example, a patient with 1 PLN out of 20 ELNs (LND = 0.05) has a better prognosis than a patient with 1 PLN out of 5 ELNs (LND = 0.2). Although both patients had the same N stage, the latter were more likely to have occult lymph nodes. In contrast to the ELN, it can only be applied for prognostic evaluation of node-positive patients at intermediate and terminal stages, and it is considered an unfavorable prognostic factor; the larger the ratio is, the poorer the survival. Based on the LNR, the weighted lymph node ratio (ωLNR) and the log odds of positive lymph nodes (LODDS) were derived as similar assessment tools^95,96:

[ω L N R = \frac{(# o f p o s i t i v e L N s) + β}{(t o t a l # o f L N s e v a l u a t e d) + β}]

[L O D D S = \log \frac{(# o f p o s i t i v e L N s) + β}{(t o t a l # o f L N s e v a l u a t e d - # o f p o s i t i v e L N s) + β}]

where β is a weighted value. The maximum value of the consistency index of the univariate Cox proportional hazards model is generally used as its value, and some researchers directly set it at 0.5.^95,96 One study compared the ability of the N stage, LNR, LODDS, and ELN count to predict the prognosis of patients with pancreatic cancer by using the Acme Information Criterion, receiver operating characteristic area under the curve, and the consistency index. They ultimately found that the LNR had the strongest predictive power.⁹⁷

It is important to note that the LNR cannot be used as a prognostic factor independent of the PLN count. Researchers have argued that from the perspective of surgery and anatomy, the same LNR does not necessarily indicate that the extent of surgery and lymph node metastasis are similar. There might even be significant differences.⁶⁶ To put it more conservatively, the LNR is not reliable without knowing the exact PLN count.

Machine Learning-Based Model

The greatest advantage of machine learning is its ability to help doctors make better decisions through the use of large amounts of digitized medical information. In addition to being used for preoperative auxiliary analysis of occult lymph nodes, machine learning is also being used for postoperative auxiliary analysis of the residual risk of occult lymph nodes. Research in this area is still limited, and the relevant literature is focused on oral cancer.^98-100 Patients included were also those at early stages without clinical evidence of lymph node metastasis who had undergone primary tumor resection, but the outcome indicator was pathologically confirmed lymph node metastasis within 2 years. Clinical variables included age, sex, race and ethnicity, body mass index, smoking history, and primary tumor site; pathological variables included the maximum size of the tumor, depth of invasion, muscle invasion, submucosal invasion, histological grade, involvement of margins, perineural invasion, and lympho-vascular invasion. In addition to clinical-pathological data, some researchers also included gene expression profiles, ie, the mRNA expression levels of genes most relevant to lymph node metastasis.¹⁰¹ By inputting the above feature variables into training, the researchers constructed supervised machine learning models based on different classification algorithms and selected the models that best matched the real situation and predicted lymph node metastasis. Such models are also promising in both reducing the number of pathologically node-negative patients undergoing unnecessary cervical lymph node dissection and identifying those at high risk of occult lymph nodes. As Farrokhian et al reported, the model correctly spared 1 patient from END for every 10.6 patients screened and correctly identified 1 additional patient with occult lymph nodes for every 21.0 patients screened compared with the traditional depth of invasion model.⁹⁸ In addition, Xu et al combined preoperative PET/CT variables with postoperative pathological variables to construct a nomogram for the prediction of the risk of occult lymph nodes, and the results demonstrated that the combination showed better predictive performance than the application of either alone.¹⁰⁰ However, they directly adopted the logistic regression algorithm without explaining why they did not choose other algorithms or comparing the pros and cons with other algorithms.

Postoperative Immunotherapy

There have been earlier clinical trials and systematic reviews that yielded conflicting results. Some have suggested that SND only prolongs surgery and increases bleeding without increasing postoperative hospitalization, complications, or decreasing survival rates¹⁰²; some have suggested that harvesting a greater number of lymph nodes is either not associated with survival or is harmful to overall survival.^103-105 A recent study performed a Restricted Cubic Splines analysis and demonstrated that the correlation between the ELN count and hazard ratio was not linear, as expected, but rather a U-shaped, dose‒response and nonlinear correlation, suggesting that excessive lymphadenectomy is actually detrimental to survival.¹⁰⁶ The reasons for the increased hazard ratio might be, on the one hand, as suggested by Jiang et al, that patients with so many lymph nodes dissected were in advanced stages. Despite eliminating lymph node micrometastasis, the risk posed by potential distant metastasis has surpassed that posed by lymph node micrometastasis. This is corroborated by other literature: extended lymph node dissection provides a survival benefit by eliminating stage migration in early-stage patients but does not improve survival in advanced-stage patients.¹⁰⁷ On the other hand, excessive lymphadenectomy can cause immune impairment (Figure 2). This opinion was first proposed by Liang et al in 2023, who followed their previous clinical study in 2017, inspired by basic medical research on the impact of removing TdLNs on immune infiltration and immunotherapy.²⁹ Basic studies have confirmed the adverse effects of TdLN removal on immune infiltration and immunotherapy, as well as the importance of TdLNs in the antitumor immune response.^108-111

Figure 2.

Balance of lymph node dissection. While less dissection may miss occult lymph nodes, excessive dissection is at risk of causing immune impairment.

Currently, immunotherapy primarily focuses on immune checkpoint blockade (ICB), with anti-PD-1/PD-L1 and anti-CTLA-4 monoclonal antibodies being the most common.^112-114 The principle of ICB is to enhance endogenous anti-cancer T cell metabolism by blocking inhibitory signals in the metabolic reprogramming process of T cells, thus supporting the energy demand, biosynthesis, and cellular signaling processes needed for effector activities. Some reviews have indicated that CD8 memory T cells in the TdLN can be activated and function upon administration of ICBs.^115-117 Through transformation to the Th1 phenotype, high levels of TdLN CD8 central memory T cells/effector memory T cells are associated with better ICB efficacy. Besides, CD8 T cells are also related to oncolytic virus therapy in immunotherapy. Kaufman et al described that oncolytic viruses achieve tumor elimination by inducing systemic innate immunity and tumor-specific adaptive immune responses.¹¹⁸ Following oncolytic cell death, tumor cells release tumor-associated antigens, pathogen-associated molecular patterns (PAMPs) and additional cellular danger-associated molecular pattern signals and cytokines, which promote the maturation of antigen-presenting cells (APCs). APCs subsequently activate antigen-specific CD4+and CD8 T cell responses. Once activated, CD8 T cells can expand into cytotoxic effector cells with the ability to travel through lymphatic flow to established tumor growth sites, where they mediate anti- tumor immunity upon antigen recognition.

In the latest study by liang et al, the included NSCLC patients were from multiple domestic institutions who experienced tumor recurrence during follow-up and received PD-1 immunotherapy.²⁹ They reproduced the locally weighted regression fitting curves and still identified 16 as the structural break point. To explain the immunological mechanism of immune impairment, bulk RNA sequencing of the ELNs was performed. xCell and CIBERSORT algorithms were applied to assess immune cell type and abundance. Based on the observed prolonged recurrence-free survival, they mapped the immune memory cell landscape and hypothesized that this phenomenon may be attributed to the presence of long-lived memory cells within the ELNs. The memory cells that predominantly influenced the efficacy of immunotherapy were ultimately designated CD8 central memory T cells based on the results of survival analysis and immune infiltration. It is consistent with previous theories about the influence of CD8 memory T cells on immunotherapy.

Conclusion

The management of lymph nodes in cancer patients involves preoperative examination, intraoperative lymphadenectomy and postoperative immunotherapy. It is a multi-modal, multi-omics, multi-stage (3 M) evaluation protocol. Preoperatively, the combination of radiomics can detect occult lymph nodes more efficiently and accurately than manual inspection. With further development of convolutional neural network architectures for image recognition, we can foresee deep learning models that achieve even higher specificity and sensitivity. Intraoperatively, we recommend that surgeons take END and SND into consideration at all times. The development of lymph node dissection will focus on individualization and precision. SND certainly allows for accurate staging, but it does not always result in a favorable outcome. For END, various guidelines are updating the minimum number of ELNs for each type of tumor, known as the threshold of ELN count. Postoperative immunotherapy is necessary because preoperative examination or intraoperative lymphadenectomy cannot detect all PLNs at 100%. Inevitable missed cases require complementary adjuvant therapy. The paradox is that less dissection may result in more residual PLNs, and more dissection may cause immunotherapy failure, which seems to result in a poor survival outcome. We therefore suggest that until there is more evidence from randomized clinical trials that extended lymph node dissection does have a therapeutic benefit, lymphadenectomy should be performed conservatively by the threshold of ELN count.

In addition, there are no validity evaluations, algorithm comparisons, or goodness-of-fit studies of various statistical methods. This review presents a longitudinal review of various evaluation protocols but does not include a cross-sectional comparison. In the future, in addition to developing more evaluation protocols, comparative methodological studies between protocols should be carried out before they are put into clinical practice to identify the best protocols suitable for various races, regions and types of cancers.

Footnotes

Abbreviations

Declaration of Conflicting Interests

The authors declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.

Funding

The author(s) disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: This work was supported by the National Natural Science Foundation of China (Nos. 81872196 and 81672690).

ORCID iDs

Ruochong Wang

Mengyun Zhao

References

Nitti

Marchet

Olivieri

, et al. Ratio between metastatic and examined lymph nodes is an independent prognostic factor after D2 resection for gastric cancer: Analysis of a large European monoinstitutional experience. Ann Surg Oncol. Nov 2003;10(9):1077-1085. doi:https://doi.org/10.1245/aso.2003.03.520

Forum

. National voluntary consensus standards for quality of cancer care. 2009:C18-C26.

Compton

Greene

. The staging of colorectal cancer: 2004 and beyond. CA Cancer J Clin. Nov-Dec 2004;54(6):295-308. doi:https://doi.org/10.3322/canjclin.54.6.295

Fielding

Arsenault

Chapuis

, et al. Clinicopathological staging for colorectal cancer: An international documentation system (IDS) and an international comprehensive anatomical terminology (ICAT). J Gastroenterol Hepatol. Jul-Aug 1991;6(4):325-344. doi:https://doi.org/10.1111/j.1440-1746.1991.tb00867.x

Guan

Jiao

Wen

, et al. Optimal examined lymph node number for accurate staging and long-term survival in rectal cancer: A population-based study. Int J Surg. Aug 1 2023;109(8):2241-2248. doi:https://doi.org/10.1097/js9.0000000000000320

Japanese Gastric Cancer A. Japanese Classification of gastric carcinoma - 2nd English edition. Gastric Cancer. Dec 1998;1(1):10-24. doi:https://doi.org/10.1007/s101209800016

Amin

Edge

Greene

, et al. AJCC cancer staging manual, vol 1024. Springer; 2017.

Okajima

Komatsu

Ichikawa

, et al. Prognostic impact of the number of retrieved lymph nodes in patients with gastric cancer. J Gastroenterol Hepatol. Sep 2016;31(9):1566-1571. doi:https://doi.org/10.1111/jgh.13306

Sura

Robertson

Kabolizadeh

. How many lymph nodes are enough?-defining the extent of lymph node dissection in stage I-III gastric cancer using the national cancer database. J Gastrointest Oncol. Dec 2018;9(6):1168-1175. doi:https://doi.org/10.21037/jgo.2018.09.07

10.

Asamura

Chansky

Crowley

, et al. The international association for the study of lung cancer lung cancer staging project: Proposals for the revision of the N descriptors in the forthcoming 8th edition of the TNM classification for lung cancer. J Thorac Oncol. Dec 2015;10(12):1675-1684. doi:https://doi.org/10.1097/jto.0000000000000678

11.

De Leyn

Dooms

Kuzdzal

, et al. Revised ESTS guidelines for preoperative mediastinal lymph node staging for non-small-cell lung cancer. Eur J Cardiothorac Surg. May 2014;45(5):787-798. doi:https://doi.org/10.1093/ejcts/ezu028

12.

Liang

Shen

, et al. Impact of examined lymph node count on precise staging and long-term survival of resected non-small-cell lung cancer: A population study of the US SEER database and a Chinese multi-institutional registry. J Clin Oncol. Apr 10 2017;35(11):1162-1170. doi:https://doi.org/10.1200/jco.2016.67.5140

13.

Zhu

Song

Jiao

, et al. A large real-world cohort study of examined lymph node standards for adequate nodal staging in early non-small cell lung cancer. Transl Lung Cancer Res. Feb 2021;10(2):815-825. doi:https://doi.org/10.21037/tlcr-20-1024

14.

Divi

Chen

Nussenbaum

, et al. Lymph node count from neck dissection predicts mortality in head and neck cancer. J Clin Oncol. Nov 10 2016;34(32):3892-3897. doi:https://doi.org/10.1200/jco.2016.67.3863

15.

Ebrahimi

Clark

Amit

, et al. Minimum nodal yield in oral squamous cell carcinoma: Defining the standard of care in a multicenter international pooled validation study. Ann Surg Oncol. 2014;21(9):3049-3055. doi:https://doi.org/10.1245/s10434-014-3702-x

16.

Kuo

Mehra

Sosa

, et al. Proposing prognostic thresholds for lymph node yield in clinically lymph node-negative and lymph node-positive cancers of the oral cavity. Cancer. Dec 1 2016;122(23):3624-3631. doi:https://doi.org/10.1002/cncr.30227

17.

Lee

Kim

Cha

Nam

. Prognostic value of lymph node count from selective neck dissection in oral squamous cell carcinoma. Int J Oral Maxillofac Surg. Aug 2018;47(8):953-958. doi:https://doi.org/10.1016/j.ijom.2018.03.007

18.

Agrama

Reiter

Cunnane

Topham

Keane

. Nodal yield in neck dissection and the likelihood of metastases. Otolaryngol Head Neck Surg. Feb 2003;128(2):185-190. doi:https://doi.org/10.1067/mhn.2003.67

19.

Zhuang

Chen

Liu

. Valuation of lymph node dissection in localized high-risk renal cell cancer using X-tile software. Int Urol Nephrol. Feb 2020;52(2):253-262. doi:https://doi.org/10.1007/s11255-019-02307-x

20.

Ducreux

Cuhna

Caramella

, et al. Cancer of the pancreas: ESMO clinical practice guidelines for diagnosis, treatment and follow-up. Ann Oncol. Sep 2015;26(Suppl 5):v56-v68. doi:https://doi.org/10.1093/annonc/mdv295

21.

Huang

Jansen

Balavarca

, et al. Significance of examined lymph node number in accurate staging and long-term survival in resected stage I-II pancreatic cancer-more is better? A large international population-based cohort study. Ann Surg. Dec 1 2021;274(6):e554-e563. doi:https://doi.org/10.1097/sla.0000000000003558

22.

Tian

Yang

, et al. The minimum number of examined lymph nodes for accurate nodal staging and optimal survival of stage T1-2 esophageal squamous cell carcinoma: A retrospective multicenter cohort with SEER database validation. Int J Surg. Aug 2022;104:106764. doi:https://doi.org/10.1016/j.ijsu.2022.106764

23.

Huang

, et al. The prognostic impact of occult lymph node metastasis in node-negative gastric cancer: A systematic review and meta-analysis. Ann Surg Oncol. Nov 2013;20(12):3927-3934. doi:https://doi.org/10.1245/s10434-013-3021-7

24.

Wang

, et al. Predicting occult lymph node metastasis by nomogram in patients with lung adenocarcinoma ≤2 cm. Future Oncol. Jun 2021;17(16):2005-2013. doi:https://doi.org/10.2217/fon-2020-0905

25.

Moon

Choi

Park

Lee

. Risk factors for occult lymph node metastasis in peripheral non-small cell lung cancer with invasive component size 3 cm or less. World J Surg. May 2020;44(5):1658-1665. doi:https://doi.org/10.1007/s00268-019-05365-5

26.

Liefers

Cleton-Jansen

van de Velde

, et al. Micrometastases and survival in stage II colorectal cancer. N Engl J Med. Jul 23 1998;339(4):223-228. doi:https://doi.org/10.1056/nejm199807233390403

27.

Jeuck

Wittekind

. Gastric carcinoma: Stage migration by immunohistochemically detected lymph node micrometastases. Gastric Cancer. Jan 2015;18(1):100-108. doi:https://doi.org/10.1007/s10120-014-0352-4

28.

Feinstein

Sosin

Wells

. The will rogers phenomenon. Stage migration and new diagnostic techniques as a source of misleading statistics for survival in cancer. N Engl J Med. Jun 20 1985;312(25):1604-1608. doi:https://doi.org/10.1056/nejm198506203122504

29.

Deng

Zhou

Chen

, et al. Impact of lymphadenectomy extent on immunotherapy efficacy in post-resectional recurred non-small cell lung cancer: A multi-institutional retrospective cohort study. Int J Surg. Sep 27 2023;110(1):238-252. doi:https://doi.org/10.1097/js9.0000000000000774

30.

Whitson

Groth

Maddaus

. Recommendations for optimal use of imaging studies to clinically stage mediastinal lymph nodes in non-small-cell lung cancer patients. Lung Cancer. Aug 2008;61(2):177-185. doi:https://doi.org/10.1016/j.lungcan.2007.12.019

31.

Zhang

Shao

. PET/CT for predicting occult lymph node metastasis in gastric cancer. Curr Oncol. Sep 11 2022;29(9):6523-6539. doi:https://doi.org/10.3390/curroncol29090513

32.

Zhang

, et al. Diagnostic value of fluorine 18 fluorodeoxyglucose positron emission tomography/computed tomography for the detection of metastases in non-small-cell lung cancer patients. Int J Cancer. Jan 15 2013;132(2):E37-E47. doi:https://doi.org/10.1002/ijc.27779

33.

Liao

Hsu

Wang

Lai

. Detection of cervical lymph node metastasis in head and neck cancer patients with clinically N0 neck-a meta-analysis comparing different imaging modalities. BMC Cancer. Jun 12 2012;12:236. doi:https://doi.org/10.1186/1471-2407-12-236

34.

Liao

Chen

Liang

Yeh

Kao

. Meta-analysis study of lymph node staging by 18 F-FDG PET/CT scan in non-small cell lung cancer: Comparison of TB and non-TB endemic regions. Eur J Radiol. Nov 2012;81(11):3518-3523. doi:https://doi.org/10.1016/j.ejrad.2012.02.007

35.

Hofman

Smeeton

Rankin

Nunan

O'Doherty

. Observer variation in FDG PET-CT for staging of non-small-cell lung carcinoma. Eur J Nucl Med Mol Imaging. Feb 2009;36(2):194-199. doi:https://doi.org/10.1007/s00259-008-0946-3

36.

Kirmani

Rintoul

Win

, et al. Stage migration: Results of lymph node dissection in the era of modern imaging and invasive staging for lung cancer. Eur J Cardiothorac Surg. Jan 2013;43(1):104-109. ; discussion 109-10. doi:10.1093/ejcts/ezs184.

37.

Darling

Maziak

Inculet

, et al. Positron emission tomography-computed tomography compared with invasive mediastinal staging in non-small cell lung cancer: Results of mediastinal staging in the early lung positron emission tomography trial. J Thorac Oncol. Aug 2011;6(8):1367-1372. doi:https://doi.org/10.1097/JTO.0b013e318220c912

38.

Gómez-Caro

Boada

Cabañas

, et al. False-negative rate after positron emission tomography/computer tomography scan for mediastinal staging in cI stage non-small-cell lung cancer. Eur J Cardiothorac Surg. Jul 2012;42(1):93-100; discussion 100. doi:https://doi.org/10.1093/ejcts/ezr272

39.

Beyaz

Verhoeven

RLJ

Schuurbiers

OCJ

Verhagen

van der Heijden

. Occult lymph node metastases in clinical N0/N1 NSCLC; A single center in-depth analysis. Lung Cancer. Dec 2020;150:186-194. doi:https://doi.org/10.1016/j.lungcan.2020.10.022

40.

Stahl

Ott

Weber

, et al. FDG PET imaging of locally advanced gastric carcinomas: Correlation with endoscopic and histopathological findings. Eur J Nucl Med Mol Imaging. Feb 2003;30(2):288-295. doi:https://doi.org/10.1007/s00259-002-1029-5

41.

de Bree

Takes

Castelijns

, et al. Advances in diagnostic modalities to detect occult lymph node metastases in head and neck squamous cell carcinoma. Head Neck. Dec 2015;37(12):1829-1839. doi:https://doi.org/10.1002/hed.23814

42.

Lambin

Rios-Velazquez

Leijenaar

, et al. Radiomics: Extracting more information from medical images using advanced feature analysis. Eur J Cancer. Mar 2012;48(4):441-446. doi:https://doi.org/10.1016/j.ejca.2011.11.036

43.

Jiang

Yuan

, et al. Radiomic signature of (18)F fluorodeoxyglucose PET/CT for prediction of gastric cancer survival and chemotherapeutic benefits. Theranostics. 2018;8(21):5915-5928. doi:https://doi.org/10.7150/thno.28018

44.

Jiang

Wang

Chen

, et al. Radiomics signature on computed tomography imaging: Association with lymph node metastasis in patients with gastric cancer. Front Oncol. 2019;9:340. doi:https://doi.org/10.3389/fonc.2019.00340

45.

Wallis

Soussan

Lacroix

Akl

Duboucher

Buvat

. An [18F]FDG-PET/CT deep learning method for fully automated detection of pathological mediastinal lymph nodes in lung cancer patients. Eur J Nucl Med Mol Imaging. Feb 2022;49(3):881-888. doi:https://doi.org/10.1007/s00259-021-05513-x

46.

Guglielmo

Marturano

Bettinelli

, et al. Additional Value of PET and CT Image-Based Features in the Detection of Occult Lymph Node Metastases in Lung Cancer: A Systematic Review of the Literature. Diagnostics (Basel). Jun 23 2023;13(13). doi:https://doi.org/10.3390/diagnostics13132153

47.

Qiao

Zhang

Wang

Xin

. (18)F-FDG PET/CT radiomics nomogram for predicting occult lymph node metastasis of non-small cell lung cancer. Front Oncol. 2022;12:974934. doi:https://doi.org/10.3389/fonc.2022.974934

48.

Ouyang

Zheng

Wang

, et al. Deep learning analysis using (18)F-FDG PET/CT to predict occult lymph node metastasis in patients with clinical N0 lung adenocarcinoma. Front Oncol. 2022;12:915871. doi:https://doi.org/10.3389/fonc.2022.915871

49.

Zhong

Cai

Chen

, et al. PET/CT based cross-modal deep learning signature to predict occult nodal metastasis in lung cancer. Nat Commun. Nov 18 2023;14(1):7513. doi:https://doi.org/10.1038/s41467-023-42811-4

50.

Fawcett

. An introduction to ROC analysis. Pattern Recognit Lett. 2006;27(8):861-874.

51.

Sun

Mutasa

Liu

, et al. Deep learning prediction of axillary lymph node status using ultrasound images. Comput Biol Med. Apr 2022;143:105250. doi:https://doi.org/10.1016/j.compbiomed.2022.105250

52.

Zhang

Zhao

Zhang

, et al. Prediction of axillary lymph node metastatic load of breast cancer based on ultrasound deep learning radiomics nomogram. Technol Cancer Res Treat. Jan-Dec 2023;22:15330338231166218. doi:https://doi.org/10.1177/15330338231166218

53.

Wang

Zhao

Wan

, et al. Prediction of sentinel lymph node metastasis in breast cancer by using deep learning radiomics based on ultrasound images. Medicine (Baltimore). Nov 3 2023;102(44):e35868. doi:https://doi.org/10.1097/md.0000000000035868

54.

Wang

Zhang

, et al. Development of a deep learning model to identify lymph node metastasis on magnetic resonance imaging in patients with cervical cancer. JAMA Netw Open. Jul 1 2020;3(7):e2011625. doi:https://doi.org/10.1001/jamanetworkopen.2020.11625

55.

Yang

Zheng

. Computer-aided diagnostic models to classify lymph node metastasis and lymphoma involvement in enlarged cervical lymph nodes using PET/CT. Med Phys. Jan 2023;50(1):152-162. doi:https://doi.org/10.1002/mp.15901

56.

Borgemeester

van den Brekel

van Tinteren

, et al. Ultrasound-guided aspiration cytology for the assessment of the clinically N0 neck: Factors influencing its accuracy. Head Neck. Nov 2008;30(11):1505-1513. doi:https://doi.org/10.1002/hed.20903

57.

Castelijns

van den Brekel

. Imaging of lymphadenopathy in the neck. Eur Radiol. Apr 2002;12(4):727-738. doi:https://doi.org/10.1007/s003300101102

58.

van den Brekel

Castelijns

Stel

Golding

Meyer

Snow

. Modern imaging techniques and ultrasound-guided aspiration cytology for the assessment of neck node metastases: A prospective comparative study. Eur Arch Otorhinolaryngol. 1993;250(1):11-17. doi:https://doi.org/10.1007/bf00176941

59.

Stoeckli

Haerle

Strobel

Haile

Hany

Schuknecht

. Initial staging of the neck in head and neck squamous cell carcinoma: A comparison of CT, PET/CT, and ultrasound-guided fine-needle aspiration cytology. Head Neck. Apr 2012;34(4):469-476. doi:https://doi.org/10.1002/hed.21764

60.

Blancas

García-Puche

Bermejo

, et al. Low number of examined lymph nodes in node-negative breast cancer patients is an adverse prognostic factor. Ann Oncol. Nov 2006;17(11):1644-1649. doi:https://doi.org/10.1093/annonc/mdl169

61.

Caplin

Cerottini

Bosman

Constanda

Givel

. For patients with Dukes’ B (TNM stage II) colorectal carcinoma, examination of six or fewer lymph nodes is related to poor prognosis. Cancer. Aug 15 1998;83(4):666-672.

62.

Tepper

O'Connell

Niedzwiecki

, et al. Impact of number of nodes retrieved on outcome in patients with rectal cancer. J Clin Oncol. Jan 1 2001;19(1):157-163. doi:https://doi.org/10.1200/jco.2001.19.1.157

63.

Gajra

Newman

Gamble

Kohman

Graziano

. Effect of number of lymph nodes sampled on outcome in patients with stage I non–small-cell lung cancer. J Clin Oncol. 2003;21(6):1029-1034. doi:https://doi.org/10.1200/jco.2003.07.010

64.

Smith

Schwarz

. 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. Oct 1 2005;23(28):7114-7124. doi:https://doi.org/10.1200/jco.2005.14.621

65.

Voyer

TEL

Sigurdson

Hanlon

, et al. Colon cancer survival is associated with increasing number of lymph nodes analyzed: A secondary survey of intergroup trial INT-0089. J Clin Oncol. 2003;21(15):2912-2919. doi:https://doi.org/10.1200/jco.2003.05.062

66.

Shrime

Bachar

Lea

, et al. Nodal ratio as an independent predictor of survival in squamous cell carcinoma of the oral cavity. Head Neck. Nov 2009;31(11):1482-1488. doi:https://doi.org/10.1002/hed.21114

67.

Fisher

Bauer

Wickerham

, et al. Relation of number of positive axillary nodes to the prognosis of patients with primary breast cancer. An NSABP update. Cancer. Nov 1 1983;52(9):1551-1557.

68.

Chirieac

Suehiro

Niemisto

, et al. Size and number of examined lymph nodes impacts outcome in patients with stage II colorectal cancer. J Clin Oncol. 2004;22(14_suppl):3507-3507. doi:https://doi.org/10.1200/jco.2004.22.90140.3507

69.

Okada

Sadahiro

Suzuki

, et al. The size of retrieved lymph nodes correlates with the number of retrieved lymph nodes and is an independent prognostic factor in patients with stage II colon cancer. Int J Colorectal Dis. Dec 2015;30(12):1685-1693. doi:https://doi.org/10.1007/s00384-015-2357-9

70.

Kiricuta

Tausch

. A mathematical model of axillary lymph node involvement based on 1446 complete axillary dissections in patients with breast carcinoma. Cancer. May 15 1992;69(10):2496-2501.

71.

Iyer

Hanlon

Fowble

, et al. Accuracy of the extent of axillary nodal positivity related to primary tumor size, number of involved nodes, and number of nodes examined. Int J Radiat Oncol Biol Phys. Jul 15 2000;47(5):1177-1183. doi:https://doi.org/10.1016/s0360-3016(00)00574-5

72.

Suzuma

Sakurai

Yoshimura

Umemura

Tamaki

Naito

. A mathematical model of axillary lymph node involvement considering lymph node size in patients with breast cancer. Breast Cancer. 2001;8(3):206-212. doi:https://doi.org/10.1007/bf02967510

73.

Joseph

Sigurdson

Hanlon

, et al. Accuracy of determining nodal negativity in colorectal cancer on the basis of the number of nodes retrieved on resection. Ann Surg Oncol. Apr 2003;10(3):213-218. doi:https://doi.org/10.1245/aso.2003.03.059

74.

Gönen

Schrag

Weiser

. Nodal staging score: A tool to assess adequate staging of node-negative colon cancer. J Clin Oncol. Dec 20 2009;27(36):6166-6171. doi:https://doi.org/10.1200/jco.2009.23.7958

75.

Zhou

Liu

, et al. Pathological Nodal Staging Score for Gastric Signet Ring Cell Carcinoma: A Clinical Tool of Adequate Nodal Staging. Diagnostics (Basel). Sep 22 2022;12(10). doi:https://doi.org/10.3390/diagnostics12102289

76.

Shariat

Rink

Ehdaie

, et al. Pathologic nodal staging score for bladder cancer: A decision tool for adjuvant therapy after radical cystectomy. Eur Urol. Feb 2013;63(2):371-378. doi:https://doi.org/10.1016/j.eururo.2012.06.008

77.

Zheng

Wang

, et al. Impact of examined lymph node count for oesophageal squamous cell carcinoma in patients who underwent left transthoracic oesophagectomy. Eur J Surg Oncol. Oct 2020;46(10 Pt A):1956-1962. doi:https://doi.org/10.1016/j.ejso.2020.04.047

78.

Zhang

Jia

, et al. Prediction of cervical lymph nodes metastasis in papillary thyroid carcinoma (PTC) using nodal staging score (NSS). J Oncol. 2022;2022:9351911. doi:https://doi.org/10.1155/2022/9351911

79.

Robinson

Thomas

Dinan

Roman

Sosa

Hyslop

. How many lymph nodes are enough? Assessing the adequacy of lymph node yield for papillary thyroid cancer. J Clin Oncol. 2016;34(28):3434-3439. doi:https://doi.org/10.1200/jco.2016.67.6437

80.

Fowble

. Postmastectomy radiation: Then and now. Oncology (Williston Park). Feb 1997;11(2):213-234, 239. discussion 239-40, 243.

81.

McCulloch

Searle

. Generalized, linear, and mixed models. John Wiley & Sons; 2004.

82.

Griffiths

. Maximum likelihood estimation for the beta-binomial distribution and an application to the household distribution of the total number of cases of a disease. Biometrics. Dec 1973;29(4):637-648.

83.

Gil

Carlson

Boyle

, et al. Lymph node density is a significant predictor of outcome in patients with oral cancer. Cancer. Dec 15 2009;115(24):5700-5710. doi:https://doi.org/10.1002/cncr.24631

84.

Huang

Zhu

, et al. Lymph node status and its impact on the prognosis of left-sided and right-sided colon cancer: A SEER population-based study. Cancer Med. Dec 2021;10(23):8708-8719. doi:https://doi.org/10.1002/cam4.4357

85.

Camp

Dolled-Filhart

Rimm

. X-tile: A new bio-informatics tool for biomarker assessment and outcome-based cut-point optimization. Clin Cancer Res. Nov 1 2004;10(21):7252-7259. doi:https://doi.org/10.1158/1078-0432.Ccr-04-0713

86.

Stein

Cai

Groshen

Skinner

. Risk factors for patients with pelvic lymph node metastases following radical cystectomy with en bloc pelvic lymphadenectomy: Concept of lymph node density. J Urol. Jul 2003;170(1):35-41. doi:https://doi.org/10.1097/01.ju.0000072422.69286.0e

87.

Ooki

Yamashita

Kobayashi

, et al. Lymph node metastasis density and growth pattern as independent prognostic factors in advanced esophageal squamous cell carcinoma. World J Surg. Nov 2007;31(11):2184-2191. doi:https://doi.org/10.1007/s00268-007-9198-9

88.

Liao

Qiu

Zheng

, et al. Lymph node density as an independent prognostic factor in node-positive renal-cell carcinoma: Results from the surveillance, epidemiology, and End results program. Clin Genitourin Cancer. Oct 2019;17(5):e968-e980. doi:https://doi.org/10.1016/j.clgc.2019.05.011

89.

Shi

Liu

Cao

Shan

Zhang

Wang

. Development and validation of lymph node ratio-based nomograms for primary duodenal adenocarcinoma after surgery. Front Oncol. 2022;12:962381. doi:https://doi.org/10.3389/fonc.2022.962381

90.

Abd-Elhay

Elhusseiny

Kamel

, et al. Negative lymph node count and lymph node ratio are associated with survival in male breast cancer. Clin Breast Cancer. Dec 2018;18(6):e1293-e1310. doi:https://doi.org/10.1016/j.clbc.2018.07.003

91.

Kim

Park

, et al. Metastatic Lymph Node Ratio for Predicting Recurrence in Medullary Thyroid Cancer. Cancers (Basel). Nov 21 2021;13(22). doi:https://doi.org/10.3390/cancers13225842

92.

Suzuki

Sasaki

Takano

, et al. Lymph node ratio as a predictor for minor salivary gland cancer in head and neck. BMC Cancer. Nov 6 2021;21(1):1186. doi:https://doi.org/10.1186/s12885-021-08877-3

93.

Yamashita

Hosoda

Ema

Watanabe

. Lymph node ratio as a novel and simple prognostic factor in advanced gastric cancer. Eur J Surg Oncol. Sep 2016;42(9):1253-1260. doi:https://doi.org/10.1016/j.ejso.2016.03.001

94.

Perez

Hansen

Burdio

, et al. Lymph node ratio nomogram-based prognostic model for resected distal cholangiocarcinoma. J Am Coll Surg. Nov 1 2022;235(5):703-712. doi:https://doi.org/10.1097/xcs.0000000000000299

95.

Farrokhian

Holcomb

Dimon

, et al. Assessing prognostic value of quantitative neck dissection quality measures in patients with clinically node-negative oral cavity squamous cell carcinoma. JAMA Otolaryngol Head Neck Surg. Oct 1 2022;148(10):947-955. doi:https://doi.org/10.1001/jamaoto.2022.2312

96.

Lee

Hung

Chen

. The prognostic ability of log odds of positive lymph nodes in oral cavity squamous cell carcinoma. Medicine (Baltimore). Jul 2015;94(27):e1069. doi:https://doi.org/10.1097/md.0000000000001069

97.

Zheng

Shi

, et al. Lymph node ratio is a superior predictor in surgically treated early-onset pancreatic cancer. Front Oncol. 2022;12:975846. doi:https://doi.org/10.3389/fonc.2022.975846

98.

Farrokhian

Holcomb

Dimon

, et al. Development and validation of machine learning models for predicting occult nodal metastasis in early-stage oral cavity squamous cell carcinoma. JAMA Netw Open. Apr 1 2022;5(4):e227226. doi:https://doi.org/10.1001/jamanetworkopen.2022.7226

99.

Bur

Holcomb

Goodwin

, et al. Machine learning to predict occult nodal metastasis in early oral squamous cell carcinoma. Oral Oncol. May 2019;92:20-25. doi:https://doi.org/10.1016/j.oraloncology.2019.03.011

100.

Peng

Feng

, et al. A prediction model of nodal metastasis in cN0 oral squamous cell carcinoma using metabolic and pathological variables. Cancer Imaging. Apr 5 2023;23(1):34. doi:https://doi.org/10.1186/s40644-023-00552-z

101.

Wang

Liu

, et al. A combined prediction model for lymph node metastasis based on a molecular panel and clinicopathological factors in oral squamous cell carcinoma. Front Oncol. 2021;11:660615. doi:https://doi.org/10.3389/fonc.2021.660615

102.

Osarogiagbon

Decker

Ballman

Wigle

Allen

Darling

. Survival implications of variation in the thoroughness of pathologic lymph node examination in American college of surgeons oncology group Z0030 (alliance). Ann Thorac Surg. Aug 2016;102(2):363-369. doi:https://doi.org/10.1016/j.athoracsur.2016.03.095

103.

Saji

Tsuboi

Yoshida

, et al. Prognostic impact of number of resected and involved lymph nodes at complete resection on survival in non-small cell lung cancer. J Thorac Oncol. Nov 2011;6(11):1865-1871. doi:https://doi.org/10.1097/JTO.0b013e31822a35c3

104.

Elshaer

Gravante

Kosmin

Riaz

Al-Bahrani

. A systematic review of the prognostic value of lymph node ratio, number of positive nodes and total nodes examined in pancreatic ductal adenocarcinoma. Ann R Coll Surg Engl. Feb 2017;99(2):101-106. doi:https://doi.org/10.1308/rcsann.2016.0340

105.

Showalter

Winter

Berger

, et al. The influence of total nodes examined, number of positive nodes, and lymph node ratio on survival after surgical resection and adjuvant chemoradiation for pancreatic cancer: A secondary analysis of RTOG 9704. Int J Radiat Oncol Biol Phys. Dec 1 2011;81(5):1328-1335. doi:https://doi.org/10.1016/j.ijrobp.2010.07.1993

106.

Jiang

Shao

, et al.

Impact of removal of lymph nodes on survival in stage I-III gastric signet-ring cell cancer: The more, the better?

Ann Surg Oncol. Feb 2024;31(2):783-791. doi:https://doi.org/10.1245/s10434-023-14590-1

107.

Vuong

Graff-Baker

Dehal

, et al. Survival analysis with extended lymphadenectomy for gastric cancer: Removing stage migration from the equation. Am Surg. Oct 1 2017;83(10):1074-1079.

108.

Fransen

Schoonderwoerd

Knopf

, et al. Tumor-draining lymph nodes are pivotal in PD-1/PD-L1 checkpoint therapy. JCI Insight. Dec 6 2018;3(23). doi:https://doi.org/10.1172/jci.insight.124507

109.

Fear

Forbes

Neeve

, et al. Tumour draining lymph node-generated CD8 T cells play a role in controlling lung metastases after a primary tumour is removed but not when adjuvant immunotherapy is used. Cancer Immunol Immunother. Nov 2021;70(11):3249-3258. doi:https://doi.org/10.1007/s00262-021-02934-3

110.

van Pul

Notohardjo

JCL

Fransen

, et al. Local delivery of low-dose anti-CTLA-4 to the melanoma lymphatic basin leads to systemic T(reg) reduction and effector T cell activation. Sci Immunol. Jul 15 2022;7(73):eabn8097. doi:https://doi.org/10.1126/sciimmunol.abn8097

111.

Cai

, et al. Intratumoral tertiary lymphoid structure (TLS) maturation is influenced by draining lymph nodes of lung cancer. J Immunother Cancer. Apr 2023;11(4). doi:https://doi.org/10.1136/jitc-2022-005539

112.

Zhang

. The history and advances in cancer immunotherapy: Understanding the characteristics of tumor-infiltrating immune cells and their therapeutic implications. Cell Mol Immunol. Aug 2020;17(8):807-821. doi:https://doi.org/10.1038/s41423-020-0488-6

113.

Madden

Rathmell

. The Complex integration of T-cell metabolism and immunotherapy. Cancer Discov. Jul 2021;11(7):1636-1643. doi:https://doi.org/10.1158/2159-8290.Cd-20-0569

114.

Szeto

Finley

. Integrative approaches to cancer immunotherapy. Trends Cancer. Jul 2019;5(7):400-410. doi:https://doi.org/10.1016/j.trecan.2019.05.010

115.

Morad

Helmink

Sharma

Wargo

. Hallmarks of response, resistance, and toxicity to immune checkpoint blockade. Cell. Oct 14 2021;184(21):5309-5337. doi:https://doi.org/10.1016/j.cell.2021.09.020

116.

Wang

Zhu

Shi

Zhu

Hou

. Tumor-draining lymph nodes: opportunities, challenges, and future directions in colorectal cancer immunotherapy. J Immunother Cancer. Jan 19 2024;12(1). doi:https://doi.org/10.1136/jitc-2023-008026

117.

Cao

, et al.

Seizing the fate of lymph nodes in immunotherapy: To preserve or not?

Cancer Lett. Apr 28 2024;588:216740. doi:https://doi.org/10.1016/j.canlet.2024.216740

118.

Kaufman

Kohlhapp

Zloza

. Oncolytic viruses: A new class of immunotherapy drugs. Nat Rev Drug Discov. Aug 30 2016;15(9):660. doi:https://doi.org/10.1038/nrd.2016.178

A 3 M Evaluation Protocol for Examining Lymph Nodes in Cancer Patients: Multi-Modal,Multi-Omics,Multi-Stage Approach

Abstract

Keywords

Background

Standards for Examined Lymph Node Counts

Occult Node Metastasis and Lymph Node Micrometastasis

Stage Migration

Immune Impairment

Preoperative Examination

Intraoperative Lymphadenectomy

Multivariate Regression Analysis of the ELN Count

Thresholds of the ELN Count

Hypergeometric Distribution-Based Model

Beta-Binomial Distribution-Based Model

Survival Data-Based Model

Lymph Node Ratio or Lymph Node Density

Machine Learning-Based Model

Postoperative Immunotherapy

Conclusion

Footnotes

Abbreviations

Declaration of Conflicting Interests

Funding

ORCID iDs

References