Accurate Staging Research Articles

FP2.1 - The accuracy of radiological staging in rectal cancer patients post-long course chemoradiotherapy treatment

Abstract Aim Accurate post Long-course chemoradiotherapy treatment (LCCRT) allows safe selection of patients to rectal preservation without delaying resection in those with residual cancer. The aim of this study was to assess the accuracy of radiological staging in these patients. Methods Demographic and treatment data from consecutive patients receiving LCCRT for rectal cancer between May 2020 to December 2021 was collected. Local radiological staging post-LCCRT treatment was compared to pathological staging. Where patients had complete radiological response (CRR) and opted for watch and wait (WW) those with demonstrable local recurrence, within 12 months, were assumed to have been under-staged. Results Of 61 patients, 37 underwent surgical resection following LCCRT and 24 watch and wait for CRR. Post-LCCRT imaging was taken at a median of 10.86 weeks (IQR, 8.86- 13.00) post completion of treatment and resections 16.57 weeks (IQR 9.71 – 23.00) post-imaging. Four WW patients were diagnosed with local recurrence at a median of 56.54 weeks of follow-up (IQR, 39.9 – 67.90). The ratio of patients radiologically staged as T0:T1:T2:T3:T4 was 15%: 0%: 28%: 46%: 10%; whilst the pathology specimens demonstrated 20%: 5%: 12%: 49%: 15% respectively. Nodal staging (N0:N1:N2) was 78%: 20%: 2% with histology showing 71%: 24%: 5% respectively. Conclusion In our low-numbered study T-staging of patients with T0 and T1 disease had highest rates of over-staging such that 2 (5%) patients for potential watch and wait were not recognised as CCR by MRI alone. Nodal staging appeared more accurate although a tendency to under-stage N0 disease was seen.

FP4.6 - A Re-audit on the completion of pathology request forms following Colorectal Cancer Resection

Abstract Aim Good Histopathology is key to accurate staging for Colorectal cancer. The Royal College of Pathologists have outlined a number of important parameters to detail on the pathology request form following colorectal cancer resection surgery including confirmation of specimen, diagram, approach, dissection plane, part of BCSP, biopsy details, pre-operative chemoradiotherapy, history of IBD/familial cancer and pre-op staging. The aim of this re-audit was to detect any improvement in our adherence to these guidelines in comparison with our 2019 audit. Method The forms for all colorectal cancer resections performed from 2019-2023 were interrogated for the various parameters using pathology software DART. Result Of 1615 cases, 320 were included. 30% mentioned the approach versus 78% in the previous audit. Diagrams were completed in 36% compared to 44% previously. Conversely, only 4.5% in the previous audit had information on preoperative therapy, as compared to the current 20%. There were marginal improvements in information on screening, familial cancer and IBD. In addition, the degree of completion showed significant difference in some parameters in the case of consultants who emphasized on correct completion of forms (e.g. 97% included type of surgery, 91% had diagrams). Conclusion It is important to complete the Colorectal cancer resection pathology forms according to recommended standards to enable the pathologist to provide accurate tumour staging and thus guide appropriate management. Disappointingly, we did not find significant improvement in our adherence to the recommended standards as compared to our previous audit. We therefore need to focus on wider scale education and ensure implementation of newly designed forms with active consultant involvement and a re-audit.

Data‐Driven Modeling and High‐Precision Tracking Control of a Soft Continuum Manipulator: Enabling Robotic Sorting of Multiwire Cables

As a new class of robots, soft continuum manipulators have attracted attention due to their flexibility and compliance. However, these characteristics create challenges for precise modeling and control. This study proposes a hybrid offline and online data‐driven scheme to achieve high‐precision tracking control of a soft continuum manipulator. First, a novel multiscale deep neural network learns the manipulator model offline. Specifically, the feature fusion module extracts highly discriminative features and captures long‐term dependencies from the temporal trajectory data. The self‐attention module strengthens the ability to represent fusion features and enhances the model prediction accuracy. Then, the learnt model is updated using multisensor data online, and the proposed controller further compensates for the updated model and enhances the tracking accuracy in the movement stage. Finally, the experimental results demonstrate a significant improvement in motion accuracy under different trajectory‐tracking scenarios (i.e., deviations of <1 mm in position and <0.8° in orientation). The example of the multiwire cable sorting proves the feasibility of the proposed scheme in high‐precision industrial applications.

Consistency between Gleason score of prostate biopsy and radical prostatectomy: predictors of increased tumour grading

Introduction. Gleason grading is one of the key parameters for determining the prostate cancer (risk group after receiving the results of a prostate biopsy in addition to the level of prostate-specific antigen (PSA). However, an important clinical issue arises from changes in this grading following radical surgery. In some cases, such discrepancies in data can lead to a shift in patient management tactics from radical surgery to active follow-up recommended for patients with low cancer risk.Objective. To assess the consistency of the results of prostate biopsy and radical prostatectomy with an assessment of possible predictors of increased tumor Gleason grading.Materials & methods. Between 2019 and 2021, 193 patients with prostate cancer were selected for treatment at the Medical Centre of Moscow State University. Of these, 139 patients were chosen for the final analysis of Gleason tumour grading consistency after biopsy and radical prostatectomy.Results. The results of prostate biopsy and radical prostatectomy were comparable in 54.7% of patients (n = 76), while the remaining 45.3% (n = 63) showed a change in Gleason gradaing after radical prostatectomy: 29.5% (n = 41) patients showed an increased gradation, and 15.8% (n = 22) — decrease. Most often, the data were consistent for patients with a Gleason score 7: 63.6% for 3 + 4 = 7 and 57.9% for 4 + 3 = 7. The highest percentage of graduation increase was observed for patients from the group 3 + 3 = 6 and amounted to 48%. Cohen's kappa value was 0.351 (p < 0.001), indicating poor consistency between the results of the two studies. The total number of biopsies was associated with an increase in Gleason score after radical prostatectomy (OR = 0.816; 95% CI = 0.680 – 0.978). For example, the odds of Gleason tumour grading decreases by 18.4% when the number of biopsy specimens per unit increasesConclusion. The concordance of Gleason grading of prostate cancer after prostate biopsy and radical prostatectomy remains poor. The only predictor of improved concordance is the biopsy count, thereby reflecting biopsy quality and encouraging clinicians to move away from sextant biopsies to systematic and targeted biopsies. Increasing the consistency of results can provide more accurate staging and classification of patients according to cancer risk, thereby influencing future management.

Comparative Performance of 68Ga-PSMA-11 PET/CT and Conventional Imaging in the Primary Staging of High-Risk Prostate Cancer Patients Who Are Candidates for Radical Prostatectomy.

This prospective study aimed to (1) compare the diagnostic performance of 68Ga-PSMA-11 PET/CT with respect to conventional imaging (computed tomography (CT) and bone scintigraphy (BS)) in the primary staging of high-risk prostate cancer (PCa) patients and (2) validate PSMA-PET/CT accuracy in pelvic nodal staging in comparison with postoperative histopathology and assess PSMA-PET/CT's impact on patient management. Sixty castration-sensitive high-risk (ISUP 4-5 and/or PSA > 20 ng/mL and/or cT3) PCa patients eligible for radical prostatectomy were enrolled (median PSA 10.10 [IQR: 6.22-17.95] ng/mL). PSMA-PET/CT, compared with CT, identified nodal (N) and/or distant metastases (M1) in 56.7% (34/60) vs. 13.3% (8/60) (p < 0.001) of patients: N + 45% vs. 13.3% (p < 0.001), M1a 11.7% vs. 1.7% (p = 0.03), M1b 23.3% vs. 1.7% (p < 0.001). Compared with BS, PSMA-PET/CT localized unknown skeletal metastases in 15% (9/60) of cases, with no false negative findings. Overall, PSMA-PET/CT led to a TNM upstaging in 45.0% (27/60) of cases, with no evidence of downstaging, resulting in a change in management in up to 28.8% (17/59) of patients. Compared with histopathology data (n = 32 patients), the per-patient accuracy of PSMA-PET/TC for detecting pelvic nodal metastases was 90.6%. Overall, the above evidence supports the use of PSMA-PET/CT in the diagnostic workup of high-risk prostate cancer staging.

Lung Cancer Staging Using Chest CT and FDG PET/CT Free-Text Reports: Comparison Among Three ChatGPT Large-Language Models and Six Human Readers of Varying Experience.

Background: Although radiology reports are commonly used for lung cancer staging, this task can be challenging given radiologists' variable reporting styles as well as reports' potentially ambiguous and/or incomplete staging-related information. Objective: To compare performance of ChatGPT large-language models (LLMs) and human readers of varying experience in lung cancer staging using chest CT and FDG PET/CT free-text reports. Methods: This retrospective study included 700 patients (mean age, 73.8±29.5 years; 509 male, 191 female) from four institutions in Korea who underwent chest CT or FDG PET/CT for non-small cell lung cancer initial staging from January, 2020 to December, 2023. Examinations' reports used a free-text format, written exclusively in English or in mixed English and Korean. Two thoracic radiologists in consensus determined the overall stage group (IA, IB, IIA, IIB, IIIA, IIIB, IIIC, IVA, IVB) for each report using the AJCC 8th-edition staging system, establishing the reference standard. Three ChatGPT models (GPT-4o, GPT-4, GPT-3.5) determined an overall stage group for each report using a script-based application programming interface, zero-shot learning, and prompt incorporating a staging system summary. Six human readers (two fellowship-trained radiologists with lesser experience than the radiologists who determined the reference standard, two fellows, two residents) also independently determined overall stage groups. GPT-4o's overall accuracy for determining the correct stage among the nine groups was compared with that of the other LLMs and human readers using McNemar tests. Results: GPT-4o had an overall staging accuracy of 74.1%, significantly better than the accuracy of GPT-4 (70.1%, p=.02), GPT-3.5 (57.4%, p<.001), and resident 2 (65.7%, p<.001); significantly worse than the accuracy of fellowship-trained radiologist 1 (82.3%, p<.001) and fellowship-trained radiologist 2 (85.4%, p<.001); and not significantly different from the accuracy of fellow 1 (77.7%, p=.09), fellow 2 (75.6%, p=.53), and resident 1 (72.3%, p=.42). Conclusions: The best-performing model, GPT-4o, showed no significant difference in staging accuracy versus fellows, but significantly worse performance versus fellowship-trained radiologists. The findings do not support use of LLMs for lung cancer staging in place of expert healthcare professionals. Clinical Impact: The findings indicate the importance of domain expertise for performing complex specialized tasks such as cancer staging.

Diagnostic Performance of MRI and FDG PET/CT for Preoperative Locoregional Staging of Colon Cancer: Systematic Review and Meta-Analysis.

Background: CT is the standard-of-care test for colon cancer (CC) preoperative locoregional staging, but has limited diagnostic performance. More accurate preoperative staging would guide selection among expanding patient-tailored treatment options. Objective: To evaluate the diagnostic performance of MRI for T and N staging and of FDG PET/CT for N staging in CC locoregional staging through systematic review. Evidence Acquisition: PubMed, Embase, and Cochrane Library were searched through December 31, 2023 for studies reporting diagnostic performance of MRI or FDG PET/CT for primary (nonrectal) CC before resection without neoadjuvant therapy using histopathology as reference. Study quality was assessed using the QUADAS-2 tool. Publication bias was assessed with Deeks' funnel plot. Primary outcomes were estimated pooled predictive values, stratified by T and N categories for MRI and N categories for PET/CT. Secondary outcomes were pooled sensitivity and specificity. Evidence Synthesis: The systematic review included 11 MRI studies (686 patients) and five PET/CT studies (408 patients). Thirteen studies had at least one risk of bias or concern of applicability. Deeks' funnel plot asymmetry indicated possible publication bias in MRI studies for differentiation of T3cd-4 from T1-3ab disease and N- from N+ disease. For MRI, for discriminating T1-2 from T3-4 disease, PPV was 64.8% (95% CI [52.9-75.5%]), and NPV was 88.9% (95% CI [82.7-93.7%]); for discriminating T1-3ab from T3cd-4 disease, PPV was 83.4% (95% CI [75.0-90.3%]), and NPV was 74.6% (95% CI [58.2-86.7%]); for discriminating T1-3 from T4 disease, PPV was 94.0% (95% CI [89.4-97.3%]), and NPV was 39.9% (95% CI [24.9-56.6%]); for discriminating N- from N+ disease, PPV was 74.9% (95% CI [69.3-80.0%]), and NPV was 53.9% (95% CI [45.3-62.0%]). For PET/CT, for discriminating N- from N+ disease, PPV was 76.4% (95% CI [67.9-85.1%]), and NPV was 68.2% (95% CI [56.8-78.6%]). Across outcomes, MRI and PET/CT exhibited pooled sensitivity of 55.1-81.4% and pooled specificity of 70.3-88.1%. Conclusion: MRI had strongest predictive performance for T1-2 and T4 disease. MRI and PET/CT had otherwise limited predictive values, sensitivity, and specificity for evaluated outcomes related T and N staging. Clinical Impact: MRI and FDG PET/CT had overall limited utility for preoperative locoregional staging in colon cancer.

Radiotracer Innovations in Breast Cancer Imaging: A Review of Recent Progress.

This review focuses on the pivotal role of radiotracers in breast cancer imaging, emphasizing their importance in accurate detection, staging, and treatment monitoring. Radiotracers, labeled with radioactive isotopes, are integral to various nuclear imaging techniques, including positron emission tomography (PET) and positron emission mammography (PEM). The most widely used radiotracer in breast cancer imaging is 18F-fluorodeoxyglucose (18F-FDG), which highlights areas of increased glucose metabolism, a hallmark of many cancer cells. This allows for the identification of primary tumors and metastatic sites and the assessment of tumor response to therapy. In addition to 18F-FDG, this review will explore newer radiotracers targeting specific receptors, such as estrogen receptors or HER2, which offer more personalized imaging options. These tracers provide valuable insights into the molecular characteristics of tumors, aiding in tailored treatment strategies. By integrating radiotracers into breast cancer management, clinicians can enhance early disease detection, monitor therapeutic efficacy, and guide interventions, ultimately improving patient outcomes. Ongoing research aimed at developing more specific and sensitive tracers will also be highlighted, underscoring their potential to advance precision medicine in breast cancer care.

Correlation of preoperative sonographic staging and postoperative histopathologic staging in patients with invasive breast cancer.

To assess the accuracy of preoperative sonographic staging in patients with primary invasive breast cancer. We retrospectively analyzed a prospectively kept service database of patients with newly diagnosed, unifocal, cT1-3, invasive breast cancer. All patients were diagnosed at a single center institution between January 2013 and December 2021. Clinical T stage was assessed preoperatively by ultrasound and correlated with the definite postoperative pathologic T stage. Demographics, clinical and pathological characteristics were collected. Factors influencing accuracy, over- and underdiagnosis of sonographic staging were analyzed with multivariable regression analysis. A total of 2478 patients were included in the analysis. Median patients' age was 65years. 1577 patients (63.6%) had clinical T1 stage, 864 (34.9%) T2 and 37 (1.5%) T3 stage. The overall accuracy of sonography and histology was 76.5% (n = 1896), overestimation was observed in 9.1% (n = 225) of all cases, while underestimation occurred in 14.4% (n = 357) of all cases. Accuracy increased when clinical tumor stage cT was higher (OR 1.23; 95% CI 1.10-1.38, p ≤ 0.001). The highest accuracy was seen for patients with T2 stage (82.8%). The accuracy was lower in Luminal B tumors compared to Luminal A tumors (OR 0.71; 95% CI 0.59-0.87, p ≤ 0.001). We could not find any association between sonographic accuracy in HER2 positive patients, and demographic characteristics, or tumor-related factors. Our unicentric study showed a high accuracy of sonography in predicting T stage, especially for tumors with clinical T2 stage. Tumor stage and biological tumor factors do affect the accuracy of sonographic staging.

A log odds of positive lymph nodes-based predictive model effectively forecasts prognosis and guides postoperative adjuvant chemotherapy duration in stage III colon cancer: a multi-center retrospective cohort study

BackgroundThe log odds of positive lymph nodes (LODDS) was considered a superior staging system to N stage in colon cancer, yet its value in determining the optimal duration of adjuvant chemotherapy for stage III colon cancer patients has not been evaluated. This study aims to assess the prognostic value of a model that combines LODDS with clinicopathological information for stage III colon cancer patients and aims to stratify these patients using the model, identifying individuals who could benefit from varying durations of adjuvant chemotherapy.MethodA total of 663 consecutive patients diagnosed with stage III colon cancer, who underwent colon tumor resection between November 2007 and June 2020 at Sun Yat-sen University Cancer Center and Longyan First Affiliated Hospital of Fujian Medical University, were enrolled in this study. Survival outcomes were analyzed using Kaplan–Meier, Cox regression. Nomograms were developed to forecast patient DFS, with the Area Under the Curve (AUC) values of time-dependent Receiver Operating Characteristic (timeROC) and calibration plots utilized to assess the accuracy and reliability of the nomograms.ResultsMultivariate analysis revealed that perineural invasion (HR = 1.776, 95% CI: 1.052–3.003, P = 0.032), poor tumor differentiation (HR = 1.638, 95% CI: 1.084–2.475, P = 0.019), and LODDS groupings of 2 and 1 (HR = 1.920, 95% CI: 1.297–2.842, P = 0.001) were independent predictors of disease-free survival (DFS) in the training cohort. Nomograms constructed from LODDS, perineural invasion, and poor tumor differentiation demonstrated robust predictive performance for 3-year and 5-year DFS in both training (3-year AUC = 0.706, 5-year AUC = 0.678) and validation cohorts (3-year AUC = 0.744, 5-year AUC = 0.762). Stratification according to this model showed that patients in the high-risk group derived significant benefit from completing 8 cycles of chemotherapy (training cohort, 82.97% vs 67.17%, P = 0.013; validation cohort, 89.49% vs 63.97%, P = 0.030).ConclusionThe prognostic model, integrating LODDS, pathological differentiation, and neural invasion, demonstrates strong predictive accuracy for stage III colon cancer prognosis. Moreover, stratification via this model offers valuable insights into optimal durations of postoperative adjuvant chemotherapy.

Cancer Care in Resource-Limited Countries: Jordan as an Example.

Jordan, a lower- to middle-income country, is relatively small, but with rapidly growing population and a challenged economy. Cancer is a growing health care problem and currently ranked second, after cardiovascular diseases, as a cause of death. Jordan's national cancer registry continues to suffer from problems mostly related to long lag time in reporting, absence of outcome data, and accurate staging. The number of new patients with cancer diagnosed in Jordan is increasing at an expected, none disturbing rate, fueled by population growth, improving life expectancy, changing population structure that hosts more older population, high rate of obesity, smoking, and lack of adequate exercise. However, age-standardized rate for cancer incidence is significantly lower than Western societies, yet, mortality rate is higher. Despite efforts, cancer is still diagnosed at more advanced stages and at younger age. The Jordan breast cancer program represents a great example of opportunistic screening that led to significant downstaging of breast cancer. Efforts to evaluate the feasibility of screening programs for colorectal and lung cancers are underway. Tremendous efforts resulted in the execution of the largest clinical cancer genetics program in the region that helps identify patients and at-risk relatives for hereditary cancers. Low-resourced countries, including Jordan, will not be able to keep up with the rapidly increasing cost of cancer care. A better access to clinical trials and moving cancer care to ambulatory settings should offset some of this cost. A cancer control program that addresses all issues of cancer care from screening and early detection, through active cost-effective treatment that assures wider access to palliative care, hospice, and survivorship programs under an expanded universal health coverage, is an urgent national health priority.

550P The impact of mismatch repair status on accuracy of clinical staging in upfront resected stage II/III rectal cancer in the Netherlands

550P The impact of mismatch repair status on accuracy of clinical staging in upfront resected stage II/III rectal cancer in the Netherlands

637. UPFRONT SURGERY IN ESOPHAGEAL CANCER: ACCURACY OF PREOPERATIVE CLINICAL STAGING

Abstract Background Treatment of locally limited esophageal cancer is still a frequently discussed topic. Especially, the inaccuracy of clinical staging in lymph node-negative T2-carcinomas has to be considered further in the treatment pathway. Aim of this study was to evaluate the accuracy of clinical staging in patients treated with esophagectomy without neoadjuvant treatment. Methods We conducted a retrospective multicenter analysis at the departments of surgery from two university clinics. All patients between January 2016 and December 2023, who underwent an esophagectomy because of an esophageal adenocarcinoma or squamous cell carcinoma without neoadjuvant treatment were included. Advanced tumor stage, T staging greater than T2, was defined as exclusion criteria. Results Fifty-three patients after upfront esophagectomy were included in the analysis. In 40 (75%) patients a gastroesophageal junction adenocarcinoma was diagnosed, and 13 (25%) patients were treated because of a squamous cell carcinoma of the esophagus. Clinical T staging was T1 or T2 and all lymph node stages were included. The endosonographic preoperative staging of invasion depth was accurate in 32 (60%) patients in comparison to the histopathological staging. In 7 (13%) cases the T-values were upstaged one level and in 5 (9%) cases one level downstaged after resection. The endosonographic preoperative staging of nodal spread was accurate in 32 (60%) patients. Furthermore, in 7 (13%) cases node status was upstaged and in 5 (9%) patients downstaged. Nine (17%) patients did not undergo a sufficient endosonographic staging, mainly due to prior endoscopic submucosal dissection or stenosis. In these patients, clinical lymph node staging was accurate in 4 patients. Conclusion Preoperative staging using endoscopic ultrasound and computed tomography is not accurate enough in a high number of patients, therefore we are seeking to identify other predictive factors underlying advanced stage esophageal cancer.

Inspection of Deep Tumor Margins for Accurate Cutaneous Squamous Cell Carcinoma Staging.

Inspection of Deep Tumor Margins for Accurate Cutaneous Squamous Cell Carcinoma Staging.

Recent advances in the diagnosis of rectal cancer and evaluation of clinical treatment efficacy using magnetic resonance imaging

Accurate preoperative tumor staging is essential for precise treatment of rectal cancer, and magnetic resonance imaging (MRI) is the preferred method. In addition to accurately determining tumor (T) and stages node (N), MRI can also be used to evaluate factors related to the prognosis of rectal cancer, such as depth of tumor invasion into the muscular layer, extramural vascular invasion of the tumor, and distance to the circumferential resection margin. MRI is also an important method for evaluating tumor regression after neoadjuvant therapy, and quantitative biomarkers of functional MRI can objectively reflect treatment efficacy. Moreover, MRI can predict response after neoadjuvant therapy, local recurrence rate, and disease-free survival before treatment. The present article reviews recent advances in the preoperative diagnosis and staging of rectal cancer, reassessment after neoadjuvant therapy, and evaluation of clinical treatment efficacy using MRI.

431. DISSECTION OF THE SUPRACARINAL MESOESOPHAGUS RECURRENTIAL LYMPHADENECTOMY AND RECURRENT NERVE DAMAGE

Abstract Background Currently, an extended mediastinal lymphadenectomy is considered paramount for an accurate cancer staging as well as to improve the oncologic outcomes of patients with locally advanced esophageal cancer. Following the recently described supracarinal mesoesophagus (Cuesta et. al. 2023) allows a complete lymphadenectomy, including the left recurrential area. However, this dissection represents a considerable challenge and could be associated with significant complications. The aim of this study is to analyse if the extended lymphadenectomy guided by the anatomy of the superior mesoesophagus enables a more careful dissection of the recurrent nerves in order to avoid an injury of these structures. Methods We prospectively studied all patients in our centre with advanced esophageal cancer, from 2018 to 2023, who underwent esophagectomy and recurrent lymphadenectomy following the surgical plane of the supracarinal mesoesophagus (both unilateral and bilateral). All procedures were performed as curative intent by the same surgeon. We analysed all patients with recurrent nerve lesions and their severity, as well as their implication through the postoperative period and the treatment needed for their repair. In addition, we also studied the number of adenopathies included in the specimen, specifying those that belonged to the recurrential area and its positivity for tumoral cells. Results We included 40 patients in our study, 30 men and 10 women with an average age of 65 years. All of them received neoadyuvant therapy previous to the surgery (mostly CROSS and FLOT schemes). There were only 2 patients (5%) who suffered from permanent recurrent palsy, both needing surgical treatment for its correction. Two other patients had mild dysphonia that could be resolved with phoniatric rehabilitation. No fatal outcome occurred secondary to recurrent nerve injury. The median number of adenopathies resected in total was 35, while the median number of adenopathies corresponding to the recurrential area was 5. Conclusions In this study we have observed that, following the landmarks of the supracarinal mesoesophagus allows us to perform an extended mediastinal lymphadenectomy reducing the recurrent nerves injury down to only 5%. Even though a lesion of the recurrent nerves could imply a major influence on the quality of life, performing this dissection meticulously could improve the quality of cancer staging, overall and disease-free survival without significantly increasing the risk of recurrent nerve damage. For this reason, the proper knowledge of these anatomical planes of dissection may help improve lymphadenectomies without increasing or even reducing the rate of recurrent nerve injury.

799. OPTIMUM NUMBERS OF DISSECTED LYMPH NODES FOR ACCURATE NODAL STAGING IN CT1-2N0 ESOPHAGEAL SQUAMOUS CELL CARCINOMA

Abstract Background Patients with esophageal squamous cell carcinoma (ESCC) with lymph node metastasis may be misclassified as pN0 due to an inadequate lymph node dissection. The purpose of this study was to confirm that patients with cT1-2N0 ESCC are indeed pN0 and to propose an adequate number of lymph nodes for correct nodal staging. Methods A total of 1,855 patients who underwent upfront surgery for cT1-2N0 ESCC at two tertiary academic hospital in Korea between 1995 and 2019 were included in the study. The probability of false-negative lymphadenectomy and NSS (nodal staging score) was calculated to assess the probability of pN0 status. Results The mean age was 63.36 ± 8.01 years, with 1,726 (93.05%) male patients. The mean number of dissected lymph nodes was 34.69 ± 13.98. The clinical T stages were as follows: cTis &amp; cT1a 354 (19.08%), cT1b 1,053 (56.77%), and cT2 448 (24.15%). Pathologic nodal upstaging occurred in 443 (23.91%); pN1 352 (19.0%), pN2 78 (4.21%), and pN3 13 (0.7%). By calculating the probability of false-negative lymphadenectomy, removing fewer than 33 lymph nodes was associated with a 20% risk of missing a positive lymph node, whereas removing 53 lymph nodes was associated with a 10% risk of missing a positive lymph node. The NSS, which reflects the probability that a patient with ypN0 status is actually free of occult nodal disease after surgery, increased as the number of dissected lymph nodes increased. Conclusions Extensive lymph node dissection is mandatory for accurate staging and reducing the risk of missing positive lymph nodes in early-stage ESCC. The survival benefits of extensive lymph node dissection need to be evaluated in further analysis.

Refining sleep staging accuracy: Transfer learning coupled with scorability models.

This study aimed to 1) improve sleep staging accuracy through transfer learning, to achieve or exceede human inter-expert agreement; 2) introduce a scorability model to assess the quality and trustworthiness of automated sleep staging. A deep neural network (base model) was trained on a large multi-site polysomnography (PSG) dataset from the United States. Transfer learning was used to calibrate the model to a reduced montage and limited samples from the Korean Genome and Epidemiology Study (KoGES) dataset. Model performance was compared to inter-expert reliability among three human experts. A scorability assessment was developed to predict the agreement between the model and human experts. Initial sleep staging by the base model showed lower agreement with experts (κ=0.55) compared to inter-expert agreement (κ=0.62). Calibration with 324 randomly sampled training cases matched expert agreement levels. Further targeted sampling improved performance, with models exceeding inter-expert agreement (κ=0.70). The scorability assessment, combining biosignal quality and model confidence features, predicted model-expert agreement moderately well (R²=0.42). Recordings with higher scorability scores demonstrated greater model-expert agreement than inter-expert agreement. Even with lower scorability scores, model performance was comparable to inter-expert agreement. Fine-tuning a pre-trained neural network through targeted transfer learning significantly enhances sleep staging performance for an atypical montage, achieving and surpassing human expert agreement levels. The introduction of a scorability assessment provides a robust measure of reliability, ensuring quality control and enhancing the practical application of the system before deployment. This approach marks an important advancement in automated sleep analysis, demonstrating the potential for AI to exceed human performance in clinical settings.

Tumour deposits are associated with worse survival than extranodal extension; a network meta-analysis on tumour nodules in colorectal cancer.

Lymph node metastases (LNM) play a central role in the tumour-node-metastasis (TNM) classification for colorectal cancer (CRC), with extranodal extension (ENE) as an adverse feature. ENE has never been directly compared to tumour deposits (TD). The aim of this study was to perform an up-to-date systematic review, including a network meta-analysis to compare their prognostic value. A comprehensive search was conducted on PubMed, Embase, Web of Science and Cochrane databases to identify all prognostic studies on ENE and TD. A total of 20 studies were included, with 7719 cases. The primary outcome was 5-year disease-free survival (DFS); secondary outcomes were overall survival (OS) and disease-specific survival (DSS). Frequentist paired and network meta-analyses were performed using the netmeta package in R. For univariable DFS analysis, LNM + TD+ cases had a significantly worse outcome compared with LNM + ENE+ cases [hazard ratio (HR) = 1.27, 95% confidence interval (CI) = 1.06-1.53], which was no longer significant for multivariable DFS analysis (HR = 1.13, 95% CI = 0.87-1.46). All OS and multivariable DSS analyses showed a significantly worse outcome for LNM + TD+ cases compared with LNM + ENE cases. For all outcomes, both LNM + TD+ and LNM + ENE+ had a significantly increased hazard compared with LNM+ cases. This study shows that there is a trend towards worse outcome for LNM + TD+ than LNM + ENE+, not statistically significant in multivariable DFS analysis. Both groups perform significantly worse than cases with LNM only. To improve the accuracy of CRC staging, we recommend to put more emphasis on both ENE and TD in the TNM classification, with the most prominent role for TD.

Model-free aperiodic tracking for discrete-time systems using hierarchical reinforcement learning

In the operation of practical systems, it is often a challenging task to deal with limited knowledge of the system dynamics, therefore, it is needed to obtain a framework for the simultaneous learning and control of dynamic systems, able to cope with uncertain dynamics. This paper proposes a model-free aperiodic tracking control method based on MAXQ hierarchical reinforcement learning for unknown dynamics in discrete-time systems. Firstly, a MAXQ hierarchical framework is developed for aperiodic tracking control that decomposes the task into pre-control and accurate control subtasks. The former achieves sub-optimal tracking which prompts the implementation of accurate control. The latter implements aperiodic tracking based on the pre-control result to reduce resource occupation. The structure of the MAXQ framework simplifies the complexity of the tracking task, and the incorporation of pre-control accelerates the learning process in the formal control stage. Secondly, considering the disparities between the control inputs and the control update instants, pre-control is decomposed to learn the triggering strategy and control policy, respectively. Meanwhile, the control policy learns optimal control inputs by minimizing the cost function. Similarly, the triggering strategy and control policy are also separately learned in accurate control. In contrast to traditional aperiodic triggering mechanisms, in accurate control stage, the triggering strategy is developed based on the cumulative error of the pre-control result, thus avoiding the influence of accidental factors and the cumulative effects of errors, enhancing system robustness. Thirdly, the proposed tracking control is derived by using only the input, output, and reference signal data from the system, without relying on system dynamics. It is applicable to both linear and nonlinear systems, demonstrating strong generalizability. Finally, simulation examples are provided to validate the effectiveness and superiority of the proposed method.