Adaptation and External Validation of Pathogenic Urine Culture Prediction in Primary Care Using Machine Learning.

BACKGROUND: Urinary tract infection (UTI) symptoms are common in primary care, but antibiotics are appropriate only when an infection is present. Urine culture is the reference standard test for infection, but results take >1 day. A machine learning predictor of urine cultures showed high accuracy for an emergency department (ED) population but required urine microscopy features that are not routinely available in primary care (the NeedMicro classifier). METHODS: We redesigned a classifier (NoMicro) that does not depend on urine microscopy and retrospectively validated it internally (ED data set) and externally (on a newly curated primary care [PC] data set) using a multicenter approach including 80,387 (ED) and 472 (PC) adults. We constructed machine learning models using extreme gradient boosting (XGBoost), artificial neural networks, and random forests (RFs). The primary outcome was pathogenic urine culture growing ≥100,000 colony forming units. Predictor variables included age; gender; dipstick urinalysis nitrites, leukocytes, clarity, glucose, protein, and blood; dysuria; abdominal pain; and history of UTI. RESULTS: Removal of microscopy features did not severely compromise performance under internal validation: NoMicro/XGBoost receiver operating characteristic area under the curve (ROC-AUC) 0.86 (95% CI, 0.86-0.87) vs NeedMicro 0.88 (95% CI, 0.87-0.88). Excellent performance in external (PC) validation was also observed: NoMicro/RF ROC-AUC 0.85 (95% CI, 0.81-0.89). Retrospective simulation suggested that NoMicro/RF can be used to safely withhold antibiotics for low-risk patients, thereby avoiding antibiotic overuse. CONCLUSIONS: The NoMicro classifier appears appropriate for PC. Prospective trials to adjudicate the balance of benefits and harms of using the NoMicro classifier are appropriate.

A retrospective study comparing outcomes in a midwestern US population after introduction of IADPSG guidelines for gestational diabetes.

OBJECTIVE: More evidence is required to endorse the 1-step approach for gestational diabetes mellitus (GDM) for clinical practice. Since 2010, our department has pragmatically allowed faculty to self-select the guidelines they use to screen and diagnose GDM. We sought to compare the maternal and neonatal outcomes from these two simultaneous cohorts. STUDY DESIGN: We performed a retrospective cohort study of all singleton pregnancies delivered between October 2011 and -November 2013 at our hospital. Patients were excluded if they had preexisting diabetes, were not screened or screened inappropriately, or their fetus had congenital anomalies. Patients were grouped by their screening strategy, and maternal and neonatal outcomes were analyzed. RESULTS: The 1-step group had a higher incidence of GDM (21.6% versus 5.0%). Initial results suggested higher rates of neonatal hypoglycemia, phototherapy for hyperbilirubinemia, and a lower rate of gestational HTN. After adjustment, these differences disappeared, but a lower rate of large for gestational age (LGA) infants was discovered (adjusted odds ratios (aOR) 0.78). CONCLUSION: The picture remains unclear as to whether the 1-step approach is associated with significantly improved outcomes compared with the 2-step approach. We did find a lower risk for a LGA infant in our 1-step cohort, but it is unlikely that the 1-step approach would be cost-effective due to the absence of other improved outcomes.

Extrapulmonary small cell carcinoma: the University of Kansas experience and review of literature.

Though extrapulmonary small cell carcinoma was first described over 80 years ago, definitive treatment recommendations are lacking. The treatment strategies commonly utilized are extrapolated from pulmonary small cell carcinoma experience. A better understanding of this entity is needed to improve management approach. The University of Kansas tumor registry was reviewed from 1990 to 2013. Thirty-five cases met the inclusion and exclusion criteria for review. Age, gender, smoking status, weight loss, metastatic disease-related data, stage, performance status (PS), treatment received, and survival data were collected. Patients were evaluated with a variety of primary locations of disease including GI tract (29%), GU tract (35%), Gyn organs (17%), head and neck (14%), and unknown primary (9%). Several sites of metastatic disease were noted, with 57 and 43% of patients meeting criteria for limited disease (LD) and extensive disease (ED), respectively. Chemotherapy, surgery, and radiation were used in several different regimens, with small cell lung cancer-type regimens incorporating a platinum and etoposide being the most common (74%). Patients with LD had a median survival of 36 months compared with 5 months in patients with ED (p<0.0001). Among different primary sites, patients with GU and Gyn LD tumors had best median survival of 36 months. Among other variables that were examined with respect to their poor prognostic significance, PS>2 (p=0.001) and one or fewer number of treatment modalities especially in LD (p=0.0005) were found to be associated significantly with mortality. GI and GU tract tumors were the most common primary sites of disease in our retrospective review. Survival varied according to stage, PS, site of primary disease, use of chemotherapy, and number of treatment modalities used. Further studies are needed to better understand this rare disorder and optimize management approach.

Leukemoid reaction and autocrine growth of bladder cancer induced by paraneoplastic production of granulocyte colony-stimulating factor--a potential neoplastic marker: a case report and review of the literature.

INTRODUCTION: Granulocyte colony-stimulating factor produced by nonhematopoietic malignant cells is able to induce a leukemoid reaction by excessive stimulation of leukocyte production. Expression of granulocyte colony-stimulating factor and its functional receptors have been confirmed in bladder cancer cells. In vitro studies have demonstrated that granulocyte colony-stimulating factor/receptor exhibits a high affinity binding and this biological axis increases proliferation of the carcinoma. Urothelial carcinoma of the bladder is rarely associated with a leukemoid reaction and autocrine growth induced by paraneoplastic production of granulocyte colony-stimulating factor. In the world literature, there have been less than 35 cases reported in the last 35 years. The clinicopathological aspects, biology, prognosis and management of granulocyte colony-stimulating factor-secreting bladder cancers are poorly understood. CASE PRESENTATION: A 39-year-old Caucasian woman with an invasive high-grade urothelial carcinoma presented with hematuria and low-grade fevers. Laboratory tests revealed an elevated white blood cell count and absolute neutrophil count and an elevated 24-hour urine protein. Upon further evaluation she was found to have locally advanced high-grade urothelial carcinoma without nodal or distant metastasis. Her serum granulocyte colony-stimulating factor level was 10 times the normal limit. This led to the diagnosis of a paraneoplastic leukemoid reaction. Her white blood cell count immediately normalized after cystectomy but increased in concordance with recurrence of her disease. Unfortunately, she rapidly progressed and expired within 10 months from the time of first diagnosis. CONCLUSIONS: This is one of the few cases reported that illustrates the existence of a distinct and highly aggressive subtype of bladder cancer which secretes granulocyte colony-stimulating factor. Patients presenting with a leukemoid reaction should be tested for granulocyte colony-stimulating factor/receptor biological axis. Moreover, granulocyte colony-stimulating factor could be a potential neoplastic marker as it can follow the clinical course of the underlying tumor and thus be useful for monitoring its evolution. Neoadjuvant chemotherapy should be considered in these patients due to the aggressive nature of these tumors. With a better understanding of the biology, this autocrine growth signal could be a potential target for therapy in future.