Use of Artificial Intelligence Chatbots in Interpretation of Pathology Reports

Importance

Anatomic pathology reports are an essential part of health care, containing vital diagnostic and prognostic information. Currently, most patients have access to their test results online. However, the reports are complex and are generally incomprehensible to laypeople. Artificial intelligence chatbots could potentially simplify pathology reports.

Objective

To evaluate the ability of large language model chatbots to accurately explain pathology reports to patients.

Design, Setting, and Participants

This cross-sectional study used 1134 pathology reports from January 1, 2018, to May 31, 2023, from a multispecialty hospital in Brooklyn, New York. A new chat was started for each report, and both chatbots (Bard [Google Inc], hereinafter chatbot 1; GPT-4 [OpenAI], hereinafter chatbot 2) were asked in sequential prompts to explain the reports in simple terms and identify key information. Chatbot responses were generated between June 1 and August 31, 2023. The mean readability scores of the original and simplified reports were compared. Two reviewers independently screened and flagged reports with potential errors. Three pathologists reviewed the flagged reports and categorized them as medically correct, partially medically correct, or medically incorrect; they also recorded any instances of hallucinations.

Main Outcomes and Measures

Outcomes included improved mean readability scores and a medically accurate interpretation.

Results

For the 1134 reports included, the Flesch-Kincaid grade level decreased from a mean of 13.19 (95% CI, 12.98-13.41) to 8.17 (95% CI, 8.08-8.25; t = 45.29; P < .001) by chatbot 1 and 7.45 (95% CI, 7.35-7.54; t = 49.69; P < .001) by chatbot 2. The Flesch Reading Ease score was increased from a mean of 10.32 (95% CI, 8.69-11.96) to 61.32 (95% CI, 60.80-61.84; t = -63.19; P < .001) by chatbot 1 and 70.80 (95% CI, 70.32-71.28; t = -74.61; P < .001) by chatbot 2. Chatbot 1 interpreted 993 reports (87.57%) correctly, 102 (8.99%) partially correctly, and 39 (3.44%) incorrectly; chatbot 2 interpreted 1105 reports (97.44%) correctly, 24 (2.12%) partially correctly, and 5 (0.44%) incorrectly. Chatbot 1 had 32 instances of hallucinations (2.82%), while chatbot 2 had 3 (0.26%).

Conclusions and Relevance

The findings of this cross-sectional study suggest that artificial intelligence chatbots were able to simplify pathology reports. However, some inaccuracies and hallucinations occurred. Simplified reports should be reviewed by clinicians before distribution to patients.

Intravascular Large B-cell Lymphoma Diagnosed by Nasal Biopsy in a Patient Presenting with Bilateral Ptosis and Ophthalmoplegia

Intravascular large B-cell lymphoma (IVLBCL) is a rare type of lymphoma, involving the lumen of predominantly small blood vessels, especially capillaries. The orbit is an uncommon site of involvement for IVLBCL, and diagnosis before autopsy is even more rare as most cases are established post-mortem. Herein, the authors describe a 73-year-old male who presented with 3 weeks of progressive bilateral ptosis and ophthalmoplegia. Computed tomography (CT) and subsequent magnetic resonance imaging (MRI) revealed diffuse abnormal thickening and enhancement of bilateral orbital apices, superior orbital fissures, and cavernous sinus, along with persistent focal opacification of the left frontal and ethmoid sinuses. Infectious and inflammatory workup of serum and cerebrospinal fluid was negative. Ethmoidal sinus and middle turbinate biopsy confirmed intravascular large B-cell lymphoma and the patient was started on R-CHOP chemotherapy regimen.

Muscle-invasive bladder cancer following radiotherapy for prostate adenocarcinoma

We present a case of muscle-invasive bladder cancer arising in a man who had previously been diagnosed with prostate cancer. His prostate cancer diagnosis and subsequent treatment with external beam radiation therapy occurred over 20 years prior to being diagnosed with bladder cancer. Biopsies of the bladder mass revealed a high-grade urothelial carcinoma with prostatic invasion and MRI showed significant concern for invasion into the pelvic floor. Metastatic lesions on his right rib and left clavicle were discovered on bone scan.

Genomic Mismatch at LIMS1 Locus and Kidney Allograft Rejection

BACKGROUND

In the context of kidney transplantation, genomic incompatibilities between donor and recipient may lead to allosensitization against new antigens. We hypothesized that recessive inheritance of gene-disrupting variants may represent a risk factor for allograft rejection.

METHODS

We performed a two-stage genetic association study of kidney allograft rejection. In the first stage, we performed a recessive association screen of 50 common gene-intersecting deletion polymorphisms in a cohort of kidney transplant recipients. In the second stage, we replicated our findings in three independent cohorts of donor-recipient pairs. We defined genomic collision as a specific donor-recipient genotype combination in which a recipient who was homozygous for a gene-intersecting deletion received a transplant from a nonhomozygous donor. Identification of alloantibodies was performed with the use of protein arrays, enzyme-linked immunosorbent assays, and Western blot analyses.

RESULTS

In the discovery cohort, which included 705 recipients, we found a significant association with allograft rejection at the LIMS1 locus represented by rs893403 (hazard ratio with the risk genotype vs. nonrisk genotypes, 1.84; 95% confidence interval [CI], 1.35 to 2.50; P = 9.8×10^-5). This effect was replicated under the genomic-collision model in three independent cohorts involving a total of 2004 donor-recipient pairs (hazard ratio, 1.55; 95% CI, 1.25 to 1.93; P = 6.5×10^-5). In the combined analysis (discovery cohort plus replication cohorts), the risk genotype was associated with a higher risk of rejection than the nonrisk genotype (hazard ratio, 1.63; 95% CI, 1.37 to 1.95; P = 4.7×10^-8). We identified a specific antibody response against LIMS1, a kidney-expressed protein encoded within the collision locus. The response involved predominantly IgG2 and IgG3 antibody subclasses.

CONCLUSIONS

We found that the LIMS1 locus appeared to encode a minor histocompatibility antigen. Genomic collision at this locus was associated with rejection of the kidney allograft and with production of anti-LIMS1 IgG2 and IgG3. (Funded by the Columbia University Transplant Center and others.).

Induction regimen and survival in simultaneous heart-kidney transplant recipients

BACKGROUND

Induction therapy in simultaneous heart-kidney transplantation (SHKT) is not well studied in the setting of contemporary maintenance immunosuppression consisting of tacrolimus (TAC), mycophenolic acid (MPA), and prednisone (PRED).

METHODS

We analyzed the Organ Procurement and Transplant Network registry from January 1, 2000, to March 3, 2015, for recipients of SHKT (N = 623) maintained on TAC/MPA/PRED at hospital discharge. The study cohort was further stratified into 3 groups by induction choice: induction (n = 232), rabbit anti-thymoglobulin (r-ATG; n = 204), and interleukin-2 receptor-α (n = 187) antagonists. Survival rates were estimated using the Kaplan-Meier estimator. Multivariable inverse probability weighted Cox proportional hazard regression models were used to assess hazard ratios associated with post-transplant mortality as the primary outcome. The study cohort was censored on March 4, 2016, to allow at least 1-year of follow-up.

RESULTS

During the study period, the number of SHKTs increased nearly 5-fold. The Kaplan-Meier survival curve showed superior outcomes with r-ATG compared with no induction or interleukin-2 receptor-α induction. Compared with the no-induction group, an inverse probability weighted Cox proportional hazard model showed no independent association of induction therapy with the primary outcome. In sub-group analysis, r-ATG appeared to lower mortality in sensitized patients with panel reactive antibody of 10% or higher (hazard ratio, 0.19; 95% confidence interval, 0.05-0.71).

CONCLUSION

r-ATG may provide a survival benefit in SHKT, especially in sensitized patients maintained on TAC/MPA/PRED at hospital discharge.

Kidney Diseases Associated With Alternative Complement Pathway Dysregulation and Potential Treatment Options

Atypical hemolytic uremic syndrome and C3 glomerulopathy (dense deposit disease and C3 glomerulonephritis) are characterized as inappropriate activation of the alternative complement pathway. Genetic mutations affecting the alternative complement pathway regulating proteins (complement factor H, I, membrane cofactor protein and complement factor H-related proteins) and triggers (such as infection, surgery, pregnancy and autoimmune disease flares) result in the clinical manifestation of these diseases. A decade ago, prognosis of these disease states was quite poor, with most patients developing end-stage renal disease. Furthermore, renal transplantation in these conditions was associated with poor outcomes due to graft loss to recurrent disease. Recent advances in targeted complement inhibitor therapy resulted in significant improvement in disease remission, renal recovery, health-related quality of life and allograft survival.