Can physician judgment enhance model trustworthiness? A case study on predicting pathological lymph nodes in rectal cancer.

Explainability is key to enhancing the trustworthiness of artificial intelligence in medicine. However, there exists a significant gap between physicians' expectations for model explainability and the actual behavior of these models. This gap arises from the absence of a consensus on a physician-centered evaluation framework, which is needed to quantitatively assess the practical benefits that effective explainability should offer practitioners. Here, we hypothesize that superior attention maps, as a mechanism of model explanation, should align with the information that physicians focus on, potentially reducing prediction uncertainty and increasing model reliability. We employed a multimodal transformer to predict lymph node metastasis of rectal cancer using clinical data and magnetic resonance imaging. We explored how well attention maps, visualized through a state-of-the-art technique, can achieve agreement with physician understanding. Subsequently, we compared two distinct approaches for estimating uncertainty: a standalone estimation using only the variance of prediction probability, and a human-in-the-loop estimation that considers both the variance of prediction probability and the quantified agreement. Our findings revealed no significant advantage of the human-in-the-loop approach over the standalone one. In conclusion, this case study did not confirm the anticipated benefit of the explanation in enhancing model reliability. Superficial explanations could do more harm than good by misleading physicians into relying on uncertain predictions, suggesting that the current state of attention mechanisms should not be overestimated in the context of model explainability.

Long-term Outcome After Surgical Resection of Paraaortic Lymph Node Metastasis of Colorectal Cancer: A Multicenter Retrospective Study.

BACKGROUND: The significance of resection of paraaortic lymph node metastasis in colorectal cancer is controversial. OBJECTIVE: To clarify the prognosis of colorectal cancer after paraaortic lymph node metastasis resection. DESIGN: Multicenter retrospective study. SETTINGS: Thirty-six institutions in Japan participated in this study. PATIENTS: Patients with resected and pathologically proven paraaortic lymph node metastasis of CRC between 2010 and 2015. DATA SOURCES: Database and medical records at each institution. MAIN OUTCOME MEASURES: Overall survival after paraaortic lymph node metastasis resection, recurrence-free survival, and recurrence patterns after R0 resection of paraaortic lymph node metastasis. RESULTS: A total of 133 patients were included in the primary analysis population in this study. The 5-year overall survival rate (95% confidence interval [CI]) was 41.0% (32.0, 49.8), and the median survival (95% CI) was 4.1 (3.4, 4.7) years. Independent prognostic factors for overall survival were the pathological T stage (pT4 vs. pT1- 3, adjusted hazard ratio [aHR]: 1.91, p = 0.006), other organ metastasis (present vs. absent, aHR: 1.98, p = 0.005), time to metastases (synchronous vs. metachronous, aHR: 2.02, p = 0.02), and number of paraaortic lymph node metastasis (≥3 vs. <3, aHR: 2.13, p = 0.001). The 5-year recurrence-free survival rate (95% CI) was 21.1% (13.5, 29.7), with a median (95% CI) of 1.2 (0.9, 1.4) years. The primary tumor location (left- vs. right-sided colon, aHR: 4.77, p = 0.01; rectum vs. right-sided colon, aHR: 5.27, p = 0.006), other organ metastasis (present vs. absent, aHR: 1.90, p = 0.03), number of paraaortic lymph node metastasis (≥3 vs. <3, aHR: 2.20, p = 0.001), and hospital volume (<10 vs. ≥10, aHR: 2.18, p = 0.02) were identified as independent prognostic factors for recurrence-free survival. Paraaortic lymph node recurrence was the most common at 33.3%. LIMITATIONS: Selection bias cannot be ruled out because of the retrospective nature of the study. CONCLUSIONS: Less than three paraaortic lymph node metastasis was a favorable prognostic factor for both overall survival and recurrence-free survival. However, paraaortic lymph node metastases were considered to be a systemic disease and the significance of resection was limited. See Video Abstract.

Short- and long-term outcomes of surgical treatment for inguinal lymph node metastasis in rectal and anal canal adenocarcinoma.

AIM: The significance of lymphadenectomy and its indications in patients with inguinal lymph node metastasis (ILNM) of anorectal adenocarcinoma is unclear. This study aimed to clarify the surgical outcomes and prognostic factors of inguinal lymphadenectomy for ILNM. METHOD: This study included patients who underwent surgical resection for ILNM of rectal or anal canal adenocarcinoma with pathologically positive metastases between 1997 and 2011 at 20 participating centres in the Study Group for Inguinal Lymph Node Metastasis from Colorectal Cancer organized by the Japanese Society for Cancer of the Colon and Rectum. Clinicopathological characteristics and short- and long-term postoperative outcomes were retrospectively analysed. RESULTS: In total, 107 patients were included. The primary tumour was in the rectum in 57 patients (53.3%) and in the anal canal in 50 (46.7%). The median number of ILNMs was 2.34. Postoperative complications of Clavien-Dindo Grade III or higher were observed in five patients. The 5-year overall survival rate was 38.8%. Multivariate analysis identified undifferentiated histological type (P < 0.001), pathological venous invasion (P = 0.01) and pathological primary tumour depth T0-2 (P = 0.01) as independent prognostic factors for poor overall survival. CONCLUSION: The 5-year overall survival after inguinal lymph node dissection was acceptable, and it warrants consideration in more patients. Further larger-scale studies are needed in order to clarify the surgical indications.

Laparoscopic resection of a solitary fibrous tumor in the mesentery of the small intestine: a case report.

Solitary fibrous tumors (SFTs) are rare mesenchymal neoplasms that are usually localized in the mesothelium of soft tissues. Although SFTs of pleural origin are common, SFTs arising in the small intestine are extremely rare, and there are few reports of laparoscopic resection of these tumors. A 74-year-old woman presented to her local physician with intermittent pain in the lower abdomen. Computed tomography showed a 25 mm mass in the ileum with extramural protrusion and small intestine capsule endoscopy showed a submucosal tumor-like elevation covered by normal mucosa. The diagnosis was ileal tumor, which was removed by laparoscopic partial resection of the small intestine. Macroscopically, the tumor was found to be a substantial mass within subplasmalemmal fatty tissue that had no continuity with the muscular layer. Histological analysis showed proliferation of homogeneous spindle-shaped cells against a background of fibrous stroma. Immunostaining was positive for STAT6 and negative for KIT, Dog1, and S100, and SFT was diagnosed. The tumor was low risk according to Demicco's risk classification. In conclusion, a less invasive laparoscopic procedure is preferable if the tumor can be resected completely without applying excessive external force that results in seeding of tumor cells in the abdominal cavity.

Mucosal Congestion on the First Day Following Endoscopy Predicts Anastomotic Stricture After Esophagectomy.

BACKGROUND: Anastomotic stricture is a relatively common postoperative complication after esophagectomy. Previous studies have indicated that impaired perioperative blood perfusion at the anastomosis is associated with the occurrence of stricture. Therefore, we analyzed the association between endoscopically assessed blood perfusion during the early postoperative period and anastomotic stricture. METHODS: This retrospective study evaluated patients who underwent esophagectomy at Tokyo Medical and Dental University between 2010 and 2015. The patients had undergone nasal endoscopy on the 1st and 8th postoperative days. The findings were used to evaluate blood perfusion at the anastomosis and gastric tube, which was classified based on mucosal color as ischemia (white) or congestion (blue or black). Univariate and multivariable logistic regression analyses were performed to identify risk factors for anastomotic stricture. RESULTS: The study included 197 patients and anastomotic stricture was observed in 60 patients (30.4%). The multivariable analysis revealed that postoperative gastric tube congestion was a risk factor for stricture (odds ratio [OR]: 6.440, 95% confidence interval [CI]: 2.660-15.600; p < 0.001). Lower risks of anastomotic stricture were associated with pathological stage III-IV disease (OR: 0.325, 95% CI: 0.161-0.656; p = 0.002). CONCLUSION: This study revealed that endoscopically detected congestion at the anastomosis on the first postoperative day was a risk factor for anastomotic stricture.

[Meningeal Carcinomatosis of Sigmoid Colon Cancer-A Case Report].

A 60-year-old man with sigmoid colon cancer invading the urinary bladder underwent sigmoid colectomy and partial cystectomy. He developed headache and vomiting 12 weeks after surgery, and urgently hospitalized. Contrast MRI of head and whole spinal cord inspected no abnormality. CSF examination showed elevated CA19-9 20,551 U/mL, and the cytology revealed atypical cells similar to the sigmoid colon cancer cells. He was diagnosed as meningeal carcinomatosis, and received 1 course of CAPOX plus bevacizumab chemotherapy. He died 18 weeks after the surgery.