Molecular alterations associated with improved outcome in patients with glioblastoma treated with Tumor-Treating Fields

Background

The genomic and overall biologic landscape of glioblastoma (GB) has become clearer over the past 2 decades, as predictive and prognostic biomarkers of both de novo and transformed forms of GB have been identified. The oral chemotherapeutic agent temozolomide (TMZ) has been integral to standard-of-care treatment for nearly 2 decades. More recently, the use of non-pharmacologic interventions, such as application of alternating electric fields, called Tumor-Treating Fields (TTFields), has emerged as a complementary treatment option that increases overall survival (OS) in patients with newly diagnosed GB. The genomic factors associated with improved or lack of response to TTFields are unknown.

Methods

We performed comprehensive genomic analysis of GB tumors resected from 55 patients who went on to receive treatment using TTFields, and compared results to 57 patients who received standard treatment without TTFields.

Results

We found that molecular driver alterations in NF1, and wild-type PIK3CA and epidermal growth factor receptor (EGFR), were associated with increased benefit from TTFields as measured by progression-free survival (PFS) and OS. There were no differences when stratified by TP53 status. When NF1, PIK3CA, and EGFR status were combined as a Molecular Survival Score, the combination of the 3 factors significantly correlated with improved OS and PFS in TTFields-treated patients compared to patients not treated with TTFields.

Conclusions

These results shed light on potential driver and passenger mutations in GB that can be validated as predictive biomarkers of response to TTFields treatment, and provide an objective and testable genomic-based approach to assessing response.

Family Rank: A graphical domain knowledge informed feature ranking algorithm

MOTIVATION

When designing prediction models built with many features and relatively small sample sizes, feature selection methods often overfit training data, leading to selection of irrelevant features. One way to potentially mitigate overfitting is to incorporate domain knowledge during feature selection. Here, a feature ranking algorithm called 'Family Rank' is presented in which features are ranked based on a combination of graphical domain knowledge and feature scores computed from empirical data.

RESULTS

A simulated data set is used to demonstrate a scenario in which family rank outperforms other state-of-the-art graph based ranking algorithms, decreasing the sample size needed to detect true predictors by 2 to 3-fold. An example from oncology is then used to explore a real-world application of family rank.

AVAILABILITY

An implementation of Family Rank is freely available at https://cran.r-project.org/package=FamilyRank.

SUPPLEMENTARY INFORMATION

Supplementary data are available at Bioinformatics online.

Pan-cancer analysis of RNA expression of ANGIOTENSIN-I-CONVERTING ENZYME 2 reveals high variability and possible impact on COVID-19 clinical outcomes

Patients with cancer demonstrate particularly poor outcomes from COVID-19. To provide information essential for understanding the biologic underpinnings of this association, we analyzed whole-transcriptome RNA expression data obtained from a large cohort of cancer patients to characterize expression of ACE2, TMPRSS2, and other proteases that are involved in viral attachment to and entry into target cells. We find substantial variability of expression of these factors across tumor types and identify subpopulations expressing ACE2 at very high levels. In some tumor types, especially in gastrointestinal cancers, expression of ACE2 and TMPRSS2 is highly correlated. Furthermore, we found infiltration with T-cell and natural killer (NK) cell infiltration to be particularly pronounced in ACE2-high tumors. These findings suggest that subsets of cancer patients exist with gene expression profiles that may be associated with heightened susceptibility to SARS-CoV-2 infection, in whom malignant tumors function as viral reservoir and possibly promote the frequently detrimental hyper-immune response in patients infected with this virus.

Effect of prior therapy on tumor mutational burden in NSCLC

BACKGROUND

Higher tumor mutation burden (TMB) in advanced non-small cell lung cancer (NSCLC) is associated with superior outcomes with checkpoint inhibitor therapy. Tissue samples subject to TMB analysis may be acquired after DNA-damaging therapies such as chemotherapy or radiation. The impact of these therapies on TMB results is unclear. This retrospective analysis explored differences in TMB among treatment-naïve samples and treatment-experienced samples.

METHODS

NSCLC samples that underwent molecular profiling at a CLIA-certified genomics laboratory (Caris Life Sciences, Phoenix, AZ) and had available treatment and clinical history were identified. TMB was estimated by counting all coding variants (missense, nonsense, frameshift, in-frame InDels) identified by next-generation sequencing. Exceptions were synonymous mutations and any single nucleotide polymorphisms described as germline. History was reviewed under an IRB approved protocol to determine whether patients had received cytotoxic chemotherapy or radiation therapy in the year prior to collection of the tissue subject to TMB analysis. TMB values were compared between cohorts using the Wilcoxon test. Smoking adjusted P values were calculated using the chi-squared test of deviance.

RESULTS

TMB was calculated for 970 annotated tumor specimens. Of these, 155 patients received chemotherapy and/or radiation prior to tissue collection. The median TMB was 8 mut/Mb in both the treatment-naïve and treatment-experienced cohorts. After adjusting for smoking, there was no significant difference in TMB between these cohorts (P=0.22). When analyzed separately, neither prior chemotherapy nor prior radiation therapy influenced TMB. TMB was higher when the specimen source was collected from a metastatic site compared to the primary site.

CONCLUSIONS

Prior exposure to chemotherapy or radiation therapy was not associated with a significant difference in TMB.

Population bias in somatic measurement of microsatellite instability status

Microsatellite instability (MSI) is a key secondary effect of a defective DNA mismatch repair mechanism resulting in incorrectly replicated microsatellites in many malignant tumors. Historically, MSI detection has been performed by fragment analysis (FA) on a panel of representative genomic markers. More recently, using next-generation sequencing (NGS) to analyze thousands of microsatellites has been shown to improve the robustness and sensitivity of MSI detection. However, NGS-based MSI tests can be prone to population biases if NGS results are aligned to a reference genome instead of patient-matched normal tissue. We observed an increased rate of false positives in patients of African ancestry with an NGS-based diagnostic for MSI status utilizing 7317 microsatellite loci. We then minimized this bias by training a modified calling model that utilized 2011 microsatellite loci. With these adjustments 100% (95% CI: 89.1% to 100%) of African ancestry patients in an independent validation test were called correctly using the updated model. This poses not only a significant technical improvement but also has an important clinical impact on directing immune checkpoint inhibitor therapy.

Molecular profiling of melanoma brain metastases compared to primary cutaneous melanoma and to extracranial metastases

BACKGROUND

Brain metastases are a significant cause of mortality and morbidity for patients with melanoma. We hypothesize that the development of brain metastases may be explained by molecular heterogeneity between primary cutaneous melanoma (PCM) or extracranial (ECM) and brain (MBM) melanoma metastases.

MATERIALS AND METHODS

We compared next-generation sequencing, tumor mutational burden (TMB), and immunohistochemical staining for PD-L1 expression, among 132 MBM, 745 PCM, and 1190 ECM.

RESULTS

The most common genetic alterations among MBM included: BRAF (52.4%), NRAS (26.6%), CDKN2A (23.3%), NF1 (18.9%), TP53 (18%), ARID2 (13.8%), SETD2 (11.9%), and PBRM1 (7.5%). Four genes were found with higher frequency among MBM compared to PCM or ECM: BRAF (52.4% v 40.4% v 40.9%), SETD2 (11.9% v 1.9% v 3.9%), PBRM1 (7.5% v 1.6% v 2.6%), and DICER1 (4.4% v 0.6% v 0.4%). MBM showed higher TMB (p = .04) and higher PD-L1 expression (p = .002), compared to PCM. PD-L1 expression was slightly higher among MBM compared to ECM (p = .042), but there was no difference between TMB (p = .21).

CONCLUSIONS

Our findings suggest a unique molecular profile for MBM, including higher rates of BRAF mutations, higher TMB and higher PD-L1 expression, and also implicate chromatin remodeling in the pathogenesis of MBM.

Molecular profiling of melanoma brain metastases compared to primary cutaneous melanoma and to extracranial metastases

BACKGROUND

Brain metastases are a significant cause of mortality and morbidity for patients with melanoma. We hypothesize that the development of brain metastases may be explained by molecular heterogeneity between primary cutaneous melanoma (PCM) or extracranial (ECM) and brain (MBM) melanoma metastases.

MATERIALS AND METHODS

We compared next-generation sequencing, tumor mutational burden (TMB), and immunohistochemical staining for PD-L1 expression, among 132 MBM, 745 PCM, and 1190 ECM.

RESULTS

The most common genetic alterations among MBM included: BRAF (52.4%), NRAS (26.6%), CDKN2A (23.3%), NF1 (18.9%), TP53 (18%), ARID2 (13.8%), SETD2 (11.9%), and PBRM1 (7.5%). Four genes were found with higher frequency among MBM compared to PCM or ECM: BRAF (52.4% v 40.4% v 40.9%), SETD2 (11.9% v 1.9% v 3.9%), PBRM1 (7.5% v 1.6% v 2.6%), and DICER1 (4.4% v 0.6% v 0.4%). MBM showed higher TMB (p = .04) and higher PD-L1 expression (p = .002), compared to PCM. PD-L1 expression was slightly higher among MBM compared to ECM (p = .042), but there was no difference between TMB (p = .21).

CONCLUSIONS

Our findings suggest a unique molecular profile for MBM, including higher rates of BRAF mutations, higher TMB and higher PD-L1 expression, and also implicate chromatin remodeling in the pathogenesis of MBM.

Global alterations to the choroid plexus blood-CSF barrier in amyotrophic lateral sclerosis

The choroid plexus (CP) is a highly vascularized structure located in the ventricles that forms the blood-CSF barrier (BCSFB) and separates the blood from the cerebrospinal fluid (CSF). In addition to its role as a physical barrier, the CP functions in CSF secretion, transport of nutrients into the central nervous system (CNS) and a gated point of entry of circulating immune cells into the CNS. Aging and neurodegeneration have been reported to affect CP morphology and function and increase protein leakage from blood to the CSF. Amyotrophic lateral sclerosis (ALS) is a neurodegenerative disease associated with both upper and lower motor neuron loss, as well as altered proteomic and metabolomic signatures in the CSF. The role of the BCSFB and the CP in ALS is unknown. Here we describe a transcriptomic and ultrastructural analysis of BCSFB and CP alterations in human postmortem tissues from ALS and non-neurologic disease controls. ALS-CP exhibited widespread disruptions in tight junctional components of the CP epithelial layer and vascular integrity. In addition, we detected loss of pericytes around ALS blood vessels, accompanied by activation of platelet aggregation markers vWF and Fibrinogen, reminiscent of vascular injury. To investigate the immune component of ALS-CP, we conducted a comprehensive analysis of cytokines and chemokine panels in CP lysates and found a significant down-regulation of M-CSF and V-CAM1 in ALS, as well as up-regulation of VEGF-A protein. This phenotype was accompanied by an infiltration of MERTK positive macrophages into the parenchyma of the ALS-CP when compared to controls. Taken together, we demonstrate widespread structural and functional disruptions of the BCSFB in human ALS increasing our understanding of the disease pathology and identifying potential new targets for ALS therapeutic development.

Patterns and genomic correlates of PD-L1 expression in patients with biliary tract cancers

Background

Patients with biliary tract cancer (BTC) have a dismal prognosis and limited treatment options. Given the potential for immunotherapy in patients with BTC, we studied the expression of programmed death ligand-1 (PD-L1)/programmed death-1 (PD-1) and evaluated for associated genetic alterations in patients with BTC.

Methods

By immunohistochemistry (IHC), PD-L1 (SP142 antibody; ≥2+ and/or ≥5% staining on tumor cells considered positive) and PD-1 [NAT105 antibody; ≥1+ staining of tumor infiltrating lymphocytes (TILs) considered positive] expression was studied and next-generation sequencing (NGS) was performed using Caris Life Sciences' sequencing panel of 592 genes. A total of 652 patients with BTC were included in this study: 77 extrahepatic cholangiocarcinoma (ECC), 203 gallbladder cancer (GBC), and 372 intrahepatic cholangiocarcinoma (ICC).

Results

Of the 652 tumors 8.6% were PD-L1 positive with the following distribution: GBC 12.3% (25/203), ICC 7.3% (27/372), and ECC 5.2% (4/77). There was a statistically significant increase in BRAF, BRCA2, RNF43, and TP53 mutations in PD-L1 positive group as compared to PD-L1 negative. Among other biomarkers tested, TOP2A, tumor mutational burden (TMB) high (≥17 mutations per megabase) (10.7%), and microsatellite instability high (MSI-H) (7.1%) were increased in PD-L1 positive tumors versus PD-L1 negative tumors.

Conclusions

PD-L1 expression was noted in a small percentage (8.6%) of patients with BTC. This finding suggests potential benefit of immunotherapy in this subset of patients. Furthermore, there was a statistically significant association between PD-L1 expression and certain genomic alterations (BRAF, BRCA2, RNF43, TP53) and biomarkers (TOP2A, TMB high, MSI-H), which might direct the use of rational combination strategies and clinical trial development.

Impact of prior chemotherapy or radiation therapy on tumor mutation burden in NSCLC

2627

Background: Higher non-synonymous tumor mutation burden (TMB) in non-small cell lung cancer (NSCLC) is associated with a higher likelihood of response to checkpoint inhibitors. Tissue samples subject to TMB analysis may be obtained after exposure to cytotoxic chemotherapy or radiation therapy – both of which introduce somatic mutations in DNA and can influence the number of identified mutations. The role of TMB as a potential predictive marker for immunotherapy is evolving, and the impact of prior therapy on TMB could influence interpretation. Methods: Eligible cases were from patients with confirmed NSCLC, available clinical annotation and tumor molecular profiling including TMB analysis at a CLIA-certified genomics laboratory (Caris Life Sciences, Phoenix, AZ) using the Illumina NextSeq platform. TMB was calculated using only missense mutations that had not been previously reported as germline alterations. Treatment history was obtained for each patient under an IRB approved protocol to determine whether patients had had received chemotherapy or radiation therapy in the year prior to collection of the tissue subject to TMB analysis. Data analysis was performed using the chi-square test of deviance to evaluate whether TMB was statistically significantly different between groups, correcting for smoking status. Results: Out of 1,118 patients identified, 459 cases met all eligibility criteria and were evaluated. 76 patients (17%) received either chemotherapy or radiation prior to tissue collection. Samples acquired prior to any therapy had a median TMB of 10 mut/Mb vs. 11 mut/Mb in samples acquired after any therapy. After adjusting for smoking, there was no significant difference in TMB between these cohorts (p = 0.41). Secondary pair wise analysis showed no statistically significant difference in TMB from chemotherapy-naïve and chemotherapy-treated samples (p = 0.28). The same was true for radiation (p = 0.75). Collection of clinical data is ongoing and further analysis, including additional cases will be presented. Conclusions: Though cytotoxic chemotherapy and radiation therapy can introduce somatic mutations, prior exposure to either was not associated with a significant difference in TMB.

Molecular profiling of melanoma brain metastases (MBM) compared to primary cutaneous melanoma (CM)

9565

Background: Nearly 50% of metastatic melanoma patients develop brain metastases, warranting further investigation into the biology of this event. Methods: We analyzed 132 MBM and 745 CM submitted to Caris Life Sciences from 2015-2018, using next generation sequencing of a 44 or 592 cancer-related gene panel, tumor mutational burden (TMB), and PD-L1 expression by IHC. Genomic alterations (GA), including somatic mutations or CNA, were reported. High TMB (TMB-H) was defined as ≥17 mut/Mb. Comparison of molecular profiles, including cancer-related genes and recurrently altered pathways, between tumor sites and by genomic subgroup (BRAF, NRAS, KIT, NF1), was performed using Fisher’s exact test. Results: Among 132 MBM, 72.7% were male, with median age 62 yo (range 25-83). The most common GAs among MBM were: BRAF (52.4%), NRAS (26.6%), CDKN2A (23.3%), NF1 (18.9%), TP53 (18%), ARID2 (13.8%), SETD2 (11.9%), and PBRM1 (7.5%). Compared to CM, MBM were more often TMB-H (53.7% v 38%, p = .025), with higher PD-L1 expression, using both a ≥1% (54.4% v 35.6%, p = .002) and ≥5% cut-off (32.9% v 15.9%, p = .0006). MBM showed higher rates of GAs among: SETD2 (11.9% v 1.9%, p = .0008), BRAF (52.4% v 35.6%, p = .017), PBRM1 (7.5% v 1.6%, p = .018), KRAS (4% v 1%, p = .026), CCND1 (2.9% v 0%, p = .03), and DICER1 (4.4% v 0.6%, p = .04), compared to CM. Alterations of the MAPK (87.9% v 77.8%, p = .015) and SWI/SNF (22.1% v 11.6%, p = .036) pathway were more frequent in MBM, than CM. When analyzed by genomic subgroup, BRAF+ MBM had more GAs involving the PI3K pathway (20% v 5.1%, p = .027), compared to BRAF WT MBM. NRAS+ MBM had higher PD-L1 expression at the ≥1% cutoff (66.7% v 38.6%, p = .05), but not ≥5%, compared to NRAS WT MBM. NF1+ MBM had more GAs involving the SWI/SNF (60% v 11.6%, p = .003) pathway, as opposed to NF1 WT MBM. No significant associations were seen between KIT status, TMB, PD-L1 or other pathways among MBM. Conclusions: In this cross-sectional study, MBM demonstrated higher PD-L1 expression and were more often TMB-H, compared to CM. MBM also featured more GAs involving BRAF and the MAPK pathway. We identified two novel genes, PBRM1 and SETD2, as well as recurrent alterations of the SWI/SNF pathway, supporting future studies of chromatin remodeling pathways in MBM.

Lessons learned from an osteoporosis clinical trial in frail long-term care residents

BACKGROUND

Although osteoporosis affects women of all ages, the impact is most pronounced in frail residents in long-term care. Nevertheless, few interventional trials have been performed in this population, and few data on therapeutic alternatives are available in this cohort.

PURPOSE

We describe the challenges and lessons learned in developing and carrying out a trial in frail long-term-care residents.

METHODS

The Zoledronic acid in frail Elders to STrengthen bone (ZEST) study was designed to examine the safety and efficacy of a single-dose therapy for osteoporosis in frail residents in long-term care in the Pittsburgh area. Women with osteoporosis who were 65 years of age and older and currently not on therapy were randomized in a blinded fashion to intravenous zoledronic acid or placebo. Follow-up of each participant was planned for 2 years. All participants received appropriate calcium and vitamin D supplementation.

RESULTS

Seven hundred and thirty-three contacts were made with long-term care residents of nine participating facilities. Of 252 women screened, 181 were eligible, enrolled, and randomized. Multiple barriers to research in long-term-care facilities were encountered but overcome with direct communication, information sessions, in-service trainings, and social events. Lessons learned included designing the study in a manner that avoided placing an additional burden on an already overcommitted facility staff, a two-stage consent process to separate screening from randomization, and a flexible examination schedule to accommodate residents while obtaining the necessary outcome measurements. Furthermore, a mobile unit accessible to participants containing state-of-the-art dual x-ray absorptiometry (DXA), assessment for vertebral fractures, and phlebotomy equipment allows all assessments to be performed on-site at each facility. Serious adverse events are collected from affiliated hospitals in real time with a novel electronic surveillance system.

LIMITATIONS

The major limitation is selection of outcomes that can be assessed at participating facilities and do not require transport of participants to hospitals or clinics.

CONCLUSIONS

Clinical research for osteoporosis can be successfully and safely performed with frail residents in long-term care facilities. Lessons learned from this study may inform future investigations among frail elderly residents of these facilities.

Prediction of response to alkylator-based chemotherapy in metastatic melanoma (MM) using gene expression and promoter methylation signatures

9009

Background: Temozolomide and dacarbazine (TMZ and DTIC) remain the mainstay of alkylator-based chemotherapy for MM, despite response rates of 10–15% and the absence of any impact on survival. Classification of patients according to responsiveness can guide the individualization of therapy and inform approaches to abrogate mechanisms of chemotherapy resistance. Epigenetic mechanisms play an important role in regulation of genes associated with resistance and were evaluated in tandem with gene expression profiling in biological samples from MM patients (pts) to refine our understanding of the epigenomic-genomic-phenotypic interplay. Methods: We examined promoter methylation and gene expression in tumor tissues of 21 pts with MM treated with TMZ or DTIC, using high throughput technologies (Illumina Inc). The cases were divided into responder (R) and non-responder (NR) groups based on clinical response. The data were analyzed using Prediction Analysis of Microarrays (PAM) from BRB array tools. Results: Differential promoter methylation analysis revealed that 63.6% of promoter sites were hypomethylated in tumors obtained from R pts (p<0.0001). PAM analysis of gene expression data revealed that a classifier set consisting of 82 genes was able to predict NRs from Rs with 83% sensitivity and 89% specificity. Promoter methylation profiling did not independently correlate with R status. A simultaneous analysis of the promoter methylation and gene expression values first stratified into 3 data-driven categories and then combined into a 3 by 3 matrix allowed us to identify a common gene expression/methylation signature of 15 genes that classified both NR and R groups accurately 100% of the time. Conclusions: Gene expression signatures independently predict response to chemotherapy in MM, however promoter methylation profiling alone does not. Analysis of combined gene expression and promoter methylation in a well- annotated clinical data set dichotomized according to response identified a highly predictive signature. The findings from this study are qualified by the relatively small sample size and are currently being validated in an expanded sample set. Supported in part by the ECOG Paul Carbone, MD, Fellowship Award.

No significant financial relationships to disclose.

Expression and purification of sea raven type II antifreeze protein from Drosophila melanogaster S2 cells

We present a system for the expression and purification of recombinant sea raven type II antifreeze protein, a cysteine-rich, C-type lectin-like globular protein that has proved to be a difficult target for recombinant expression and purification. The cDNAs encoding the pro- and mature forms of the sea raven protein were cloned into a modified pMT Drosophila expression vector. These constructs produced N-terminally His(6)-tagged pro- and mature forms of the type II antifreeze protein under the control of a metallothionein promoter when transfected into Drosophila melanogaster S2 cells. Upon induction of stable cell lines the two proteins were expressed at high levels and secreted into the medium. The proteins were then purified from the cell medium in a simple and rapid protocol using immobilized metal affinity chromatography and specific protease cleavage by tobacco etch virus protease. The proteins demonstrated antifreeze activity indistinguishable from that of wild-type sea raven antifreeze protein purified from serum as illustrated by ice affinity purification, ice crystal morphology, and their ability to inhibit ice crystal growth. This expression and purification system gave yields of 95 mg/L of fully active mature sea raven type II AFP and 9.6 mg/L of the proprotein. This surpasses all previous attempts to express this protein in Escherichia coli, baculovirus-infected fall armyworm cells and Pichia pastoris and will provide sufficient protein for structural analysis.

Evaluation of the effect of spinal cord injury on serum PSA levels

OBJECTIVES

Prostatic structure and secretory activity are thought to be influenced by autonomic innervation of the prostate. Prostatic denervation is especially likely in patients with spinal cord injury (SCI) at the level of the cauda equina or the conus medullaris, where the peripheral nerve supply to the prostate may be specifically damaged. This may result in changes in serum prostate-specific antigen (PSA) levels, either directly or indirectly. Therefore, we measured serum PSA levels and also studied the influence of factors such as age, catheterization, duration of SCI, urinary tract infection, and history of cystitis on serum PSA values in men with SCI.

METHODS

Serum PSA levels were determined in 79 men with SCI (age older than 40 years) using banked sera by the Abbott MEIA PSA assay. Variables such as age, catheterization, duration of SCI, urine culture results, and history of cystitis were obtained from a review of patient records. Comparisons were made with a randomly selected, non-SCI control population of 501 men, 40 to 89 years old, who underwent serum PSA determination at our institution. Statistical comparisons were performed using the Mann-Whitney U test (nonparametric), since the populations were not normally distributed. Multivariate logistic regression analysis was used to assess the correlation between the various factors and the serum PSA levels in men with SCI.

RESULTS

No statistically significant differences were found in the median serum PSA values between the SCI group and the non-SCI control population. The age-specific PSA values obtained in the SCI group were also comparable to those reported for the general population at large. Age (P <0.03) and the presence of a catheter (P <0.0002) were the only two factors that were correlated with higher serum PSA values in the SCI group by regression analysis.

CONCLUSIONS

Men with SCI tended to have serum PSA value distributions that were similar to those of the general population. However, those in the SCI group who had indwelling catheters were more likely to have higher PSA values at baseline, as were older men with SCI.

A simple comorbidity scale predicts clinical outcomes and costs in dialysis patients

PURPOSE

In a university-based dialysis program, we found that 25% of the patients accounted for 50% of the costs and 42% of the deaths. We determined whether the Charlson Comorbidity Index, a simple measure of comorbid conditions, could predict clinical outcomes and costs in these patients.

METHODS

Patients on hemodialysis or peritoneal dialysis from July 1996 to June 1998 at the University of Pittsburgh outpatient dialysis unit were studied. Comorbidity scores and outcomes were determined by reviewing the Medical Archival Retrieval System database and outpatient records.

RESULTS

Two hundred sixty-eight patients were observed for 293 patient-years. The Comorbidity Index strongly predicted admission rate (relative risk per each unit increase = 1.20; 95% confidence interval [CI]: 1.16 to 1.23, P = 0.0001), hospital days and inpatient costs (both P <0.0001), and mortality (relative risk per unit increase = 1.24, 95% CI: 1.11 to 1.39, P = 0.0002.). Age and diabetes, used in the Health Care Financing Administration dialysis capitation model, correlated poorly with outcomes.

CONCLUSIONS

The modified Charlson Comorbidity Index predicts outcomes and costs in dialysis patients. This index may be useful in determining appropriate payment for care of dialysis patients under capitated payment schemes and as a research tool to stratify dialysis patients in order to compare the outcomes of various interventions.

Percutaneous endoscopic gastrostomy reduces total hospital costs in head-injured patients

BACKGROUND

Gastrostomies provide reliable long-term enteral access in patients with traumatic brain injuries. The impact of technique of gastrostomy on total hospital cost is not known.

METHODS

A retrospective analysis of patients who sustained head trauma and required gastrostomies for long-term enteral access between 1 July 1990 and 1 July 1996 was performed.

RESULTS

The patients who received percutaneous endoscopic gastrostomies (PEG) were similar to patients who received Stamm gastrostomies (OPEN) with respect to age, injury severity score, mechanism of injury, associated injuries, complication rates, and deaths. Total hospital costs ($ x 10(3)) were lower for patients who had PEGs placed in the intensive care unit (78.2 +/- 37.4) or endoscopy suite (71.9 +/- 37.7) compared with PEGs placed in the operating room (122.4 +/- 75.7) or OPEN gastrostomies (119.8 +/- 65.1).

CONCLUSIONS

In head-injured patients, PEGs are a reliable method of obtaining long-term enteral access with a complication rate equivalent to Stamm gastrostomies. If performed in either the intensive care unit or the endoscopy suite, PEGs are associated with significantly reduced total hospital costs.

The cost of caring for end-stage kidney disease patients: an analysis based on hospital financial transaction records

The costs of care for end-stage renal disease patients continue to rise because of increased numbers of patients. Efforts to contain these costs have focused on the development of capitated payment schemes, in which all costs for the care of these patients are covered in a single payment. To determine the effect of a capitated reimbursement scheme on care of dialysis patients (both hemodialysis [HD] and peritoneal dialysis [PD]), complete financial records (all reimbursements for inpatient and outpatient care, as well as physician collections) of dialysis patients at a single medical center over 1 year were analyzed. For the period from July 1994 to July 1995, annualized cost per dialysis patient-year averaged $63,340, or 9.8% higher than the corrected estimate from the U.S. Renal Data Service (USRDS; $57,660). The "most expensive" 25% of patients engendered 44 to 48% of the total costs, and inpatient costs accounted for 37 to 40% of total costs. Nearly half of the inpatient costs resulted from only two categories (room charges and inpatient dialysis), whereas other categories each made up a small fraction of the inpatient costs. PD patients were far less expensive to care for than HD patients, due to reduced hospital days and lower cost of outpatient dialysis. Care for a university-based dialysis population was only slightly more expensive than estimates predicted from the USRDS. These results validate the USRDS spending data and suggest that they can be used effectively for setting capitated rates. Efforts to control costs without sacrificing quality of care must center on reducing inpatient costs, particularly room charges and the cost of inpatient dialysis.

Using computer modeling to help identify patient subgroups in clinical data repositories

OBJECTIVE

The ability to accurately and efficiently identify patient cases of interest in a hospital information system has many important clinical, research, educational and administrative uses. The identification of cases of interest sometimes can be difficult. This paper describes a two-stage method for searching for cases of interest.

DESIGN

First, a Boolean search is performed using coded database variables. The user classifies the retrieved cases as being of interest or not. Second, based on the user-classified cases, a computer model of the patient cases of interest is constructed. The model is then used to help locate additional cases. These cases provide an augmented training set for constructing a new computer model of the cases of interest. This cycle of modeling and user classification continues until halted by the user.

MEASUREMENTS

This paper describes a pilot study in which this method is used to identify the records of patients who have venous thrombosis.

RESULTS

The results indicate that computer modeling enhances the identification of patient cases of interest.

Resource utilization for minimally invasive direct and standard coronary artery bypass grafting

BACKGROUND

Minimally invasive direct coronary artery bypass grafting (MIDCABG) has been recently reintroduced into the cardiac surgical armamentarium for selected patients with suitable coronary anatomy. We hypothesized that MIDCABG had the potential for similar immediate results with decreased perioperative morbidity and decreased resource utilization compared with standard coronary artery bypass grafting (CABG).

METHODS

From January 1996 to August 1996, 17 MIDCABG patients were compared with 33 patients with left ventricular ejection fraction greater than 0.50 who underwent CABG with standard technique. No significant differences were observed between the two groups for preoperative variables that are known to affect cost and resource utilization. Length of stay in the hospital was 2.5 +/- 0.8 days for MIDCABG and 5.9 +/- 2 days for CABG (p < 0.0001); length of stay in the intensive care unit was 12.3 +/- 3.3 hours for MIDCABG compared to 32.3 +/- 12.6 hours for the CABG group (p < 0.0001).

RESULTS

Forty-one percent of MIDCABG patients were extubated in the operating room and 59% were discharged home on the first or second postoperative day versus none in the CABG group (p < 0.0001). Significantly less morbidity was observed in the MIDCABG group compared with CABG. Total ratio of cost-to-charge was $12,885 +/- $1,511 for MIDCABG and $21,260 +/- $5,497 for CABG (p < 0.0001), with an average savings of $8,375.

CONCLUSIONS

Minimally invasive CABG is associated with significant reduction of resource utilization and morbidity related to inital hospitalization compared with CABG.

Utilization and diagnostic yield of blood cultures in a surgical intensive care unit

OBJECTIVE

To evaluate the diagnostic yield of blood cultures obtained in a surgical intensive care unit (ICU) and to assess factors potentially influencing yield.

DESIGN

Retrospective, descriptive study.

SETTING

Surgical ICU in a university hospital.

SUBJECTS

All patients who had a blood culture obtained during their admission to the trauma/neurosurgical ICU of Presbyterian University Hospital from January 1, 1993 to December 31, 1993.

MEASUREMENTS AND MAIN RESULTS

Blood culture isolates were categorized as pathogens or contaminants and overall diagnostic yield was determined. Blood cultures were positive for pathogens in 4.6% of all culture episodes, while contaminants were isolated in 5.5% of all culture episodes. A total of 23 true bacteremias were identified in 21 patients, for an overall rate of bacteremia of 3.6 per 100 admissions (5.9 per 1,000 patient days). Concurrent antibiotics were being used at the time of blood culture in 65.3% of all culture episodes. The yield for pathogens was significantly lower (2.2%) when cultures were obtained on antibiotics compared with culture episodes obtained off antibiotics (6.4%) (p < .05). Single-set blood culture episodes were obtained in approximately 32% of all culturing episodes with the overall yield for pathogens of these culturing episodes lower (2.9%) than that of multiple-set culture episodes (5.3%) (p = NS).

CONCLUSIONS

Blood culture yield in this surgical ICU was relatively low in comparison with other published studies. The data further suggest that concurrent use of systemic antibiotics and inappropriate or excessive culturing may negatively influence blood culture yield.

The emerging role of the health information management professional in data quality and analysis of an electronic medical record system

The role of the health information management professional is changing with the advancement of the electronic medical record system (EMRS) and electronic financial database systems. The Medical ARchival System, a longitudinal EMRS developed at the University of Pittsburgh Medical Center, was used to describe the methodology that the health information management professional should use when performing data quality evaluations and data analysis of an EMRS. Specific examples of steps used to integrate databases to collect the data and generate appropriate reports and tables are also described and discussed.

Assessing depression in primary medical and psychiatric practices

A two-stage assessment of newly admitted outpatients at community mental health centers and primary medical care centers found the Center for Epidemiological Studies Depression Scale to be highly sensitive but relatively nonspecific when compared with assessments of depression by the Diagnostic Interview Schedule. Considerably higher than usual screening scores are recommended with both patient populations to improve the efficiency of the Center for Epidemiological Studies Depression Scale. The reasons for our finding that primary care clinicians underdiagnose depressive disorder while psychiatric clinicians overdiagnose it relative to the Diagnostic Interview Schedule include varied presentations by each sector's patients, differing clinical acumen, and factors affecting third-party reimbursement.

Direct gene transfer to plants

Evidence for direct, gene-mediated stable genetic transformation of plant cells of Nicotiana tabacum is presented. A selectable hybrid gene comprising the protein coding region of the Tn5 aminoglycoside phosphotransferase type II gene under control of cauliflower mosaic virus gene VI expression signals was introduced into plant protoplasts as part of an Escherichia coli plasmid. The gene was stably integrated into plant genomic DNA and constitutively expressed in selected, drug resistant, protoplast-derived cell clones. The mode of integration of the foreign gene into the plant genome resembled that observed for DNA transfection of mammalian cells. Plants regenerated from transformed cell lines were phenotypically normal and fertile, and they maintained and expressed the foreign gene throughout the development of vegetative and generative organs. Microspores, grown in anther culture, developed into resistant and sensitive haploid plantlets. Genetic crossing analysis of one of the transformed plants revealed the presence of one dominant trait for kanamycin resistance segregating in a Mendelian fashion in the F(1) generation.

Hyoscyamus muticus + Nicotiana tabacum fusion hybrids selected via auxotroph complementation

Protoplasts of the nicotinamide-deficient Hyoscyamus muticus cell line nic(-) IVH2 and of the nitrate reductase-deficient Nicotiana tabacum cell line NR(-) cnx 68 were induced to fuse. Selection for putative interspecific hybrid clones was via auxotroph complementation. Controls included tests for cross-feeding, cross inhibition, PEG-induced variation, culture-induced variation, reversion, viability, delayed selection. Protrotrophic cell lines were recovered exclusively from PEG-treated mixtures of both protoplast types. The putative hybrid clones grew independent of exogenously supplied auxins and cytokinins, and at a faster rate than either parent. The morphogenic potential of different clones varied from 'non-morphogenic' to 'regeneration of fertile plants'. Indirect evidence for the hybrid nature of the clones is provided from a) tight selection, b) hormone-independent growth, c) hybrid vigour, d) extreme morphological variation, e) isoenzyme bands from both parents, f) morphogenic potential. Definite proof for the hybrid nature was, however, provided from species-specific DNA hybridization. Chimerism could be excluded since only the large subunit of Hyoscyamus muticus ribulose bisphosphate carboxylase was found and since species-specific DNA hybridization identified clones which gave no Nicotiana tabacum signal.

Polypeptides expressed in Escherichia coli K-12 minicells by transposition elements Tn1 and Tn3

Escherichia coli K-12 minicells were employed to examine polypeptides encoded by plasmids carrying wild-type and mutant Tn1 or Tn3 transposition elements. Tn1- and Tn3-containing minicells express high levels of four transposon-specified polypeptides. Three, of molecular weights 30,000, 28,000, and 25,000, are related immunologically to beta-lactamase, the enzyme responsible for ampicillin hydrolysis. A fourth polypeptide of molecular weight 19,000 is encoded by the Tn1 or Tn3 region which spans the BamHI cleavage site. Mutant transposons which no longer produce this polypeptide transpose at higher than wild-type frequencies to give aberrant transposition products (Gill et al., J. Bacteriol. 136: 742--756, 1978; Heffron et al., Proc. Natl. Acad. Sci U.S.A. 72:3632--3627, 1975). No expression could be detected from a region of the transposons extending from the inverted repeat sequence distal to the beta-lactamase gene to more than half the distance into the Tn1 or Tn3 sequence.