The Fidelity of Artificial Intelligence to Multidisciplinary Tumor Board Recommendations for Patients with Gastric Cancer: A Retrospective Study

PURPOSE

Due to significant growth in the volume of information produced by cancer research, staying abreast of recent developments has become a challenging task. Artificial intelligence (AI) can learn, reason, and understand the enormous corpus of literature available to the scientific community. However, large-scale studies comparing the recommendations of AI and a multidisciplinary team board (MTB) in gastric cancer treatment have rarely been performed. Therefore, a retrospective real-world study was conducted to assess the level of concordance between AI and MTB treatment recommendations.

METHODS

Treatment recommendations of Watson for Oncology (WFO) and an MTB were retrospectively analyzed 322 patients with gastric cancer from January 2015 to December 2018 and the degree of agreement between them was compared. The patients were divided into concordance and non-concordance groups and factors affecting the concordance rate were analyzed.

RESULTS

The concordance rate between the AI and MTB was 86.96%. The concordance rates for each stage were 96.93% for stage I, 88.89% for stages II, 90.91% for stage III, and 45.83% for stage IV, respectively. In the multivariate analysis, age (p-value = 0.000), performance status (p-value = 0.003 for performance score 1; p-value = 0.007 for performance score 2; p-value = 0.000 for performance score 3), and stage IV (p-value = 0.017) had a significant effect on concordance between the MTB and WFO.

CONCLUSION

Factors affecting the concordance rate were age, performance status, and stage IV gastric cancer. To increase the validity of future medical AI systems for gastric cancer treatment, their supplementation with local guidelines and the ability to comprehensively understand individual patients is essential.

Usefulness and limitations of intraoperative pathological diagnosis using frozen sections for spinal cord tumors

BACKGROUND

Intraoperative pathological diagnosis has a major influence on the intra- and postoperative management of spinal cord tumors. Thus, the aim of this study was to assess the reliability of intraoperative pathological diagnosis for spinal cord lesions by comparing it with the final pathological diagnosis and to determine its usefulness and limitations.

METHOD

Three-hundred and three consecutive patients (mean age, 53.9 years) with neoplastic spinal cord lesions who underwent initial surgery between 2000 and 2021 were included. The anatomical locations of the spinal cord tumors and the implementation rate of intraoperative pathological diagnosis in each tumor type were evaluated. As the primary outcome, we determined the concordance rates between the intraoperative pathological diagnosis and the final diagnosis. When the intraoperative pathological diagnosis and final diagnosis were the same, the diagnosis was defined as a "match." Otherwise, the diagnosis was defined as a "mismatch."

RESULTS

The overall implementation rate of intraoperative pathological diagnosis was 53%, with implementation rates of 71%, 45%, 47%, and 50% for intramedullary, intradural extramedullary, extradural, and dumbbell tumors, respectively. The overall concordance rate was 87.6%, with concordance rates of 80%, 95%, 75%, and 90% for intramedullary, intradural extramedullary, extradural, and dumbbell tumors, respectively (p < 0.05). The diagnoses of ependymomas, low-grade astrocytomas, and high-grade astrocytomas was occasionally difficult among intramedullary tumors. Among intradural extramedullary tumors, differentiation between grade 1 meningioma and high-grade meningioma was difficult using intraoperative pathological diagnosis.

CONCLUSIONS

Surgeons must recognize the lower accuracy of intraoperative pathological diagnosis for intramedullary and extradural lesions and make a final decision by considering the intraoperative gross findings, preoperative clinical course, and imaging.

High inter-rater reliability of Japanese bedriddenness ranks and cognitive function scores: a hospital-based prospective observational study

BACKGROUND

The statistical validities of the official Japanese classifications of activities of daily living (ADLs), including bedriddenness ranks (BR) and cognitive function scores (CFS), have yet to be assessed. To this aim, we evaluated the ability of BR and CFS to assess ADLs using inter-rater reliability and criterion-related validity.

METHODS

New inpatients aged ≥75 years were enrolled in this hospital-based prospective observational study. BR and CFS were assessed once by an attending nurse, and then by a social worker/medical clerk. We evaluated inter-rater reliability between different professions by calculating the concordance rate, kappa coefficient, Cronbach's α, and intraclass correlation coefficient. We also estimated the relationship of the Barthel Index and Katz Index with the BR and CFS using Spearman's correlation coefficients.

RESULTS

For the 271 patients enrolled, BR at the first assessment revealed 66 normal, 10 of J1, 15 of J2, 18 of A1, 31 of A2, 37 of B1, 35 of B2, 22 of C1, and 32 of C2. The concordance rate between the two BR assessments was 68.6%, with a kappa coefficient of 0.61, Cronbach's α of 0.91, and an intraclass correlation coefficient of 0.83, thus showing good inter-rater reliability. BR was negatively correlated with the Barthel Index (r = - 0.848, p < 0.001) and Katz Index (r = - 0.820, p < 0.001), showing justifiable criterion-related validity. Meanwhile, CFS at the first assessment revealed 92 normal, 47 of 1, 19 of 2a, 30 of 2b, 60 of 3a, 8 of 3b, 8 of 4, and 0 of M. The concordance rate between the two CFS assessments was 70.1%, with a kappa coefficient of 0.62, Cronbach's α of 0.87, and an intraclass correlation coefficient 0.78, thus also showing good inter-rater reliability. CFS was negatively correlated with the Barthel Index (r = - 0.667, p < 0.001) and Katz Index (r = - 0.661, p < 0.001), showing justifiable criterion-related validity.

CONCLUSIONS

BR and CFS could be reliable and easy-to-use grading scales of ADLs in acute clinical practice or large-scale screening, with high inter-rater reliabilities among different professions and significant correlations with well-established, though complicated to use, instruments to assess ADLs.

TRIAL REGISTRATION

UMIN000041051 (2020/7/10).

Concordance of the molecular subtype classification between core needle biopsy and surgical specimen in primary breast cancer

BACKGROUND

Molecular profiling of breast cancer (BC) classifies several intrinsic subtypes based on different patterns of gene expression. Multigene assays estimate the risk of recurrence and help to select high-risk patients for adjuvant chemotherapy. However, these tests are associated with significant costs. Immunohistochemistry (IHC) offers a surrogate classification for molecular subtypes by determining estrogen (ER) and progesterone receptors (PR), human epidermal growth factor (Her2neu), as well as the proliferation marker Ki67. Core needle biopsy (CNB) is well established in BC diagnosis and allows a pre-operative assessment of biomarkers. The aim of this study was to analyze the concordance of these markers between CNB and surgical specimens to assess whether re-testing of the surgical specimen is mandatory.

MATERIALS AND METHODS

Within a 3-year period, patients with primary BC and paired samples of CNB and surgical specimens were analyzed retrospectively. Concordance rates of ER, PR, Her2neu, Ki67, and the surrogate classification for molecular subtypes were calculated using the Landis and Koch agreement grades.

RESULTS

Out of 2254 patients with primary breast cancer, 1307 paired specimens without pre-operative treatment were available for analysis Concordance rates for ER, PR, Her2neu, and Ki67 status showed substantial-to-almost perfect agreement grades (κ = 0.91, 0.75, 0.89, and 0.61, respectively). Though substantial concordance was also found for the subtype classification (κ = 0.70), the molecular subtype changed in 18.5% of patients based on the testing of the surgical specimen, mainly from luminal A-like to luminal B-like.

CONCLUSIONS

Though the concordance rates for single markers were convincing, a significant proportion of the molecular subtypes differed between CNB and the surgical specimen. Re-testing of PR and Ki67 is mandatory to ensure optimal treatment decisions. Further research is necessary to define safe, efficient, and cost-effective predictive models in adjuvant breast cancer therapy.

The role of multiparametric resonance and biopsy in prostate cancer detection: comparison with definitive histological report after laparoscopic/robotic radical prostatectomy

Purpose

Magnetic Resonance Imaging (MRI) targeted biopsy increases overall detection rates and decreases the risk of clinically insignificant PCa detection. The aim of this retrospective study is to compare concordance rates regarding side of lesion and Gleason Score at fusion targeted/systematic biopsy and MRI with the definitive histologic report of prostatectomy specimen.

Methods

115 patients underwent multiparametric (mp) MRI and successively fusion targeted/systematic biopsy. 107 patients, with a positive biopsy for PCa, further underwent laparoscopic/robotic radical prostatectomy. We compared surgical histologic report with biopsy histologic report for side of lesion and Gleason Score. We further compared PIRADS score at mpMRI with Gleason Score of both histologic reports.

Results

Concordance rate for mpMRI lesion side was 74% compared to biopsy and 52.3% compared to surgical histologic report (p < 0.0001). Fusion targeted/systematic biopsy reported a concordance rate with surgical histologic report of 67.3% for side of the lesion, while Gleason Score was concordant for 73.6% for clinically significant cancer (Gleason Score ≥ 7) (p < 0.0001). PIRADS score ≥ 3 was further associated with clinically significant cancer at surgical histologic report in 92.4% of cases (p = 0.359).

Conclusion

Multiparametric MRI of the prostate reaches a good and improvable accuracy in the detection of suspicious PCa before biopsy. A combined approach of fusion targeted and systematic biopsy could further increase the overall accuracy in PCa diagnosis, especially in biopsy-naïve patients, reaching concordance rates with definitive histologic report up to 52.3% and 85.5%.

Comparison of Detection of Epidermal Growth Factor Receptor (EFGR) Gene Mutation in Peripheral Blood Plasma (Liquid Biopsy) with Cytological Specimens in Lung Adenocarcinoma Patients

The ctDNA plasma testing is one of the methods to examine biomarkers for lung adenocarcinoma in order to detect a mutation of epidermal growth factor receptor (EGFR) gene. The advantages of ctDNA testing over tissue biopsy and lung tumor cytology include less invasive, faster result, cheaper, and minimum risk of complication for the patient. We analyzed and compare the detection of EFGR mutation in peripheral blood plasma (liquid biopsy) with cytological specimens of patients with lung adenocarcinoma. We conducted ctDNA plasma testing in 124 lung adenocarcinoma patients who visited our hospital from January to December 2018. The ctDNA testing results were compared with the results of EGFR detection from the previous cytological specimen examination. Most of the patients were males, aged 55-59 years, nonsmokers, and had stage IVA lung adenocarcinoma, with most metastasis found in the pleura. We found a correlation between EGFR prevalence with nonsmoking status and patient's age. The ctDNA plasma testing detected 27.4% common EGFR mutation and 72.6% wild-type EGFR. The figures of EGFR mutation detection from cytological specimens were 47.6% and 52.4%, respectively. Compared to cytological specimens, the EGFR mutation detection in ctDNA had a sensitivity of 48.3%, with a specificity of 90.9%, PPV of 82.35%, NPV of 66.7%, and 70.97% concordance rate. EGFR mutation with cytological specimen examination was more accurate than ctDNA.

Ultrasound scans and dual energy CT identify tendons as preferred anatomical location of MSU crystal depositions in gouty joints

OBJECTIVE

The present study was performed to localize the articular deposition of monosodium urate (MSU) crystal in joints. We compare the detection efficiencies of dual-energy CT (DECT) and ultrasound scans.

METHODS

Analyses by DECT and ultrasound were performed with 184 bilateral joints of the lower limbs of 54 consecutive gout patients. All joints were categorized into (1) knee, (2) ankle, (3) MTP1, and (4) MTP2, and sorted into those with and those without detectable MSU deposition. The comparison of the positive rate between DECT and ultrasound and the agreement was performed using the McNemar test and the Cohen's κ coefficient, respectively. Next, we listed the MSU crystal deposition as assessed by ultrasound between the DECT-positive and -negative joints according to their interior structure. We included tendons, synovia, cartilage, subcutaneous tissue, etc. RESULTS: Among all joints, the percentages with MSU crystal deposition detected by DECT (99/184, 53.8%) and ultrasound (106/184, 57.6%) were comparable (P = 0.530 > 0.05). For MTP1 (21/34, 61.8%; 12/34, 35.3%; P < 0.05) and MTP2-5 (17/34, 50.0%; 10/34, 29.4%, P < 0.05), ultrasound and DECT were more efficient, respectively. The data concordance in 46 of 50 joints (92.00%; κ = 0.769, P < 0.05) for knee; and 27 of 34 joints (79.41%; κ = 0.588, P < 0.05) for MTP2-5 and suggested that tendons were the most frequent anatomical location of MSU crystal deposition.

CONCLUSIONS

The tendons are the most frequent anatomical location of MSU crystal depositions. The concordance rate of knee joints and MTP2-5 joints shows good agreement between DECT and ultrasound depending on the location.

Bacterial flora analysis of coliforms in sewage, river water, and ground water using MALDI-TOF mass spectrometry

The aim of this study was to rapidly and effectively analyze coliforms, which are the most fundamental indicators of water quality for fecal pollution, using matrix-assisted laser desorption ionization time-of-flight mass spectrometry (MALDI-TOF MS). Coliform bacteria were isolated from municipal sewage, river water, and groundwater. For each sample, 100 isolates were determined by MALDI-TOF MS. In addition, these same 100 isolates were also identified via 16S rRNA gene sequence analysis. Obtained MALDI-TOF MS data were compared with the 16S rRNA sequencing analysis, and the validity of MALDI-TOF MS for classification of coliform bacteria was examined. The concordance rate of bacterial identification for the 100 isolates obtained by MALDI-TOF MS analysis and 16S rRNA gene sequence analysis for sewage, river water, and ground water were 96%, 74%, and 62% at the genus level, respectively. Among the sewage, river water, and ground water samples, the coliform bacterial flora were distinct. The dominant genus of coliforms in sewage, river water, and groundwater were Klebsiella spp., Enterobacter spp., and Serratia spp., respectively. We determined that MALDI-TOF MS is a rapid and accurate tool that can be used to identify coliforms. Therefore, without using conventional 16S rRNA sequencing, it is possible to rapidly and effectively classify coliforms in water using MALDI-TOF MS.

Diagnostic Accuracy on Magnetic Resonance Imaging for the Diagnosis of Osteoarthritis of the Temporomandibular Joint

OBJECTIVES

Osteoarthritis, which is also called degenerative arthritis or degenerative joint disease, is primarily a disease that results from the breakdown and loss of cartilage in joints. The purpose of this study was to investigate the diagnostic accuracy of magnetic resonance images for the diagnosis of osteoarthritis of the temporomandibular joint.

MATERIALS AND METHODS

Fifty patients (50 joints) with closed locking of the temporomandibular joint were examined with magnetic resonance imaging and then underwent arthroscopic surgery. The agreement of osteoarthritis between magnetic resonance images and arthroscopic findings was studied using the κ coefficient.

RESULTS

The incidence of osteoarthritis on magnetic resonance images (38%) was significantly lower than that in arthroscopic findings (78%). There was no significant agreement between these two findings (p=.108). The κ coefficient was 0.154.

CONCLUSION

The diagnostic accuracy of magnetic resonance images for osteoarthritis of the temporomandibular joint was low; early osteoarthritis could not be diagnosed from magnetic resonance images. Clinicians should understand that the diagnostic accuracy of osteoarthritis without arthroscopy is not always high.

Reliability of Small Biopsy Samples Compared With Resected Specimens for the Determination of Programmed Death-Ligand 1 Expression in Non--Small-Cell Lung Cancer

BACKGROUND

Several studies have assessed the expression of programmed death-ligand 1 (PD-L1) in resected surgical specimens of non-small-cell lung cancer (NSCLC). However, the expression of PD-L1 in smaller biopsy samples of advanced NSCLC has not been reported.

PATIENTS AND METHODS

A total of 79 patients with NSCLC at our institution with available biopsy samples and resected specimens were retrospectively enrolled in the present study. PD-L1 expression was assessed by immunohistochemistry and scored using the hybrid scoring method. The concordance rates for the expression of PD-L1 between the 2 samples were analyzed.

RESULTS

The pathologic stage of the patients (51 men, 28 women; median age, 68 years) was stage I in 37, stage II in 18, and stage III in 24. The diagnostic procedures included transbronchial biopsy in 59, transbronchial needle aspiration biopsy in 14, and computed tomography (CT)-guided needle biopsy in 6. The positivity rate of PD-L1 in these samples was 38.0% (27 transbronchial biopsies, 6 transbronchial needle aspiration biopsies, 3 CT-guided needle biopsies) versus 35.4% in the resected specimens. The median hybrid score was 0 (range, 0-170), and the mean score was 28.7 ± 43.4. Comparing the biopsy samples and resected specimens with a score of ≥ 1 as positive for PD-L1 staining, 6 tumors were discordant for PD-L1 expression and 73 were concordant, for a concordance rate of 92.4% and κ value of 0.8366.

CONCLUSION

PD-L1 status showed good concordance between the biopsy samples and resected specimens. These small samples, even those derived from transbronchial needle aspiration biopsies, appear adequate for the assessment of PD-L1 expression.

Diagnostic concordance rate between histologic and cytologic specimens of endobronchial ultrasound-guided transbronchial needle aspiration in lung cancer: A single institution experience

BACKGROUND

The aspirates from endobronchial ultrasound-guided transbronchial needle aspiration (EBUS-TBNA) were examined using the tissue core and cytology. We assessed that the tissue core was a more reliable specimen and attempted to analyze how many discrepancies were found between the two specimens. We investigated diagnostic concordance rate between histology and cytology.

METHODS

From January 2011 to December 2011, a total of 267 consecutive patients with lung cancer, who underwent EBUS-TBNA, were included in this retrospective study.

RESULTS

Of the 267 lung cancer patients, 207 (77.5%) were men. The median age was 65 years old. The most common pathologic type was adenocarcinoma (120, 44.9%), followed by squamous cell carcinoma (79, 29.6%). Among the 267 patients, 579 mediastinal and hilar lymph nodes were punctured. The right lower paratracheal (204, 35.2%) and subcarinal (172, 29.7%) lymph nodes were the most common nodal stations. Among the 579 lymph nodes, malignant cells were observed in 267 (46.1%) nodes, in either the histologic, cytologic or both slides: 209 in both histology and cytology, 37 in histology only, and 21 in cytology only slides. The diagnostic concordance rate between the tissue core and cytology was 90.0% (95% confidence interval CI, 87.1-92.2) (κ = 0.79).

CONCLUSIONS

The diagnostic concordance rate between the histology and cytology of EBUS-TBNA in lung cancer was 90.0% (95% CI, 87.1-92.2) (κ = 0.79).