Assessing Performance and Clinical Usefulness in Prediction Models With Survival Outcomes: Practical Guidance for Cox Proportional Hazards Models

Risk prediction models need thorough validation to assess their performance. Validation of models for survival outcomes poses challenges due to the censoring of observations and the varying time horizon at which predictions can be made. This article describes measures to evaluate predictions and the potential improvement in decision making from survival models based on Cox proportional hazards regression.

As a motivating case study, the authors consider the prediction of the composite outcome of recurrence or death (the "event") in patients with breast cancer after surgery. They developed a simple Cox regression model with 3 predictors, as in the Nottingham Prognostic Index, in 2982 women (1275 events over 5 years of follow-up) and externally validated this model in 686 women (285 events over 5 years). Improvement in performance was assessed after the addition of progesterone receptor as a prognostic biomarker.

The model predictions can be evaluated across the full range of observed follow-up times or for the event occurring by the end of a fixed time horizon of interest. The authors first discuss recommended statistical measures that evaluate model performance in terms of discrimination, calibration, or overall performance. Further, they evaluate the potential clinical utility of the model to support clinical decision making according to a net benefit measure. They provide SAS and R code to illustrate internal and external validation.

The authors recommend the proposed set of performance measures for transparent reporting of the validity of predictions from survival models.

Using machine learning to improve the accuracy of patient deterioration predictions: Mayo Clinic Early Warning Score (MC-EWS)

OBJECTIVE

We aimed to develop a model for accurate prediction of general care inpatient deterioration.

MATERIALS AND METHODS

Training and internal validation datasets were built using 2-year data from a quaternary hospital in the Midwest. Model training used gradient boosting and feature engineering (clinically relevant interactions, time-series information) to predict general care inpatient deterioration (resuscitation call, intensive care unit transfer, or rapid response team call) in 24 hours. Data from a tertiary care hospital in the Southwest were used for external validation. C-statistic, sensitivity, positive predictive value, and alert rate were calculated for different cutoffs and compared with the National Early Warning Score. Sensitivity analysis evaluated prediction of intensive care unit transfer or resuscitation call.

RESULTS

Training, internal validation, and external validation datasets included 24 500, 25 784 and 53 956 hospitalizations, respectively. The Mayo Clinic Early Warning Score (MC-EWS) demonstrated excellent discrimination in both the internal and external validation datasets (C-statistic = 0.913, 0.937, respectively), and results were consistent in the sensitivity analysis (C-statistic = 0.932 in external validation). At a sensitivity of 73%, MC-EWS would generate 0.7 alerts per day per 10 patients, 45% less than the National Early Warning Score.

DISCUSSION

Low alert rates are important for implementation of an alert system. Other early warning scores developed for the general care ward have achieved lower discrimination overall compared with MC-EWS, likely because MC-EWS includes both nursing assessments and extensive feature engineering.

CONCLUSIONS

MC-EWS achieved superior prediction of general care inpatient deterioration using sophisticated feature engineering and a machine learning approach, reducing alert rate.

Distribution and Incidence of Blood-Borne Infection among Blood Donors from Regional Transfusion Centers in Burkina Faso: A Comprehensive Study

There is a high prevalence of blood-borne infections in West Africa. This study sought to determine the seroprevalence of blood-borne infections, including hepatitis B virus (HBV), hepatitis C virus (HCV), HIV, and syphilis, in blood donors in Burkina Faso. Blood donors were recruited from 2009 to 2013 in four major cities in Burkina Faso of urban area (Ouagadougou) and rural area (Bobo Dioulasso, Fada N’Gourma, and Ouahigouya). Serology tests including hepatitis B surface antigen, anti-HCV, anti-HIV, and rapid plasma reagin test were used for screening and were confirmed with ELISA. Disease prevalence was calculated among first-time donors. Incidence and residual risk were calculated from repeat donors. There were 166,681 donors; 43,084 had ≥ 2 donations. The overall seroprevalence of HBV, HCV, HIV, and syphilis were 13.4%, 6.9%, 2.1%, and 2.4%, respectively. The incidence rates (IRs) of HBV, HCV, HIV, and syphilis infection were 2,433, 3,056, 1,121, and 1,287 per 100,000 person-years. There was lower seroprevalence of HBV and HCV in urban area than in rural area (12.9% versus 14.0%, P < 0.001; and 5.9% versus 8.0%, P < 0.001), and no difference in HIV (2.1% versus 2.1%, P = 0.25). The IRs of new HBV, HCV, HIV, and syphilis were 2.43, 3.06, 1.12, and 1.29 per 100,000 person-years, respectively. The residual risk was one per 268 donations for HBV, one per 181 donations for HCV, and one per 1,480 donations for HIV, respectively. In conclusion, this comprehensive study from four blood donation sites in Burkina Faso showed high HBV and HCV seroprevalence and incidence with high residual risk from blood donation.

The impact of arm lymphedema on healthcare utilization during long-term breast cancer survivorship: a population-based cohort study

PURPOSE

Cancer treatment-related late effects degrade survivors' quality of life, independence, and societal integration, yet may be ameliorated, or even reversed, with effective care. Unfortunately, survivors inconsistently receive this care and the impact on their healthcare utilization is unknown. We sought to estimate differences in utilization between breast cancer (BC) survivors with and without upper extremity lymphedema; a common, remediable late effect.

METHODS

We conducted a population-based, retrospective longitudinal cohort study of survivors with incident BC diagnosed from January 1, 1990, through December 31, 2010. HC utilization was characterized using the Berenson-Eggers Type of Service (BETOS) categories. Outcomes included overall healthcare utilization as well as its compartmentalization into the BETOS categories of (1) Evaluation and management, (2) Procedures, (3) Imaging, (4) Tests, (5) Durable medical equipment, (6) Physical/occupational therapy, (7) Other, and (8) Exceptions/Unclassified.

RESULTS

The cohort included 1906 subjects of which 94% (1800) had records meeting the inclusion criteria. Mean follow-up per survivor was 12.8 years (mean, 11, range 1-25 years). Analysis revealed that (1) survivors with BC-associated lymphedema used > 30% more services annually; (2) their increased utilization lessened but persisted for at least 10 years after diagnosis; and (3) this finding of increased utilization extends across all BETOS categories, is further amplified as BMI increases, and cannot be explained solely by lymphedema-directed care.

CONCLUSIONS

BC-related lymphedema appears to be an important driver of survivors' healthcare utilization and guideline-concordant activities to reduce its incidence and severity may be cost neutral or saving.

IMPLICATIONS FOR CANCER SURVIVORS

Early detection and effective management of cancer-related late effects like lymphedema may reduce survivors' healthcare needs in the decades that follow their cancer treatment.

Designing clinical trials with (restricted) mean survival time endpoint: Practical considerations

BACKGROUND/AIMS

The difference in mean survival time, which quantifies the treatment effect in terms most meaningful to patients and retains its interpretability regardless of the shape of the survival distribution or the proportionality of the treatment effect, is an alternative endpoint that could be used more often as the primary endpoint to design clinical trials. The underuse of this endpoint is due to investigators' lack of familiarity with the test comparing the mean survival times and the lack of tools to facilitate trial design with this endpoint. The aim of this article is to provide investigators with insights and software to design trials with restricted mean survival time as the primary endpoint.

METHODS

A closed-form formula for the asymptotic power of the test of restricted mean survival time difference is presented. The effects of design parameters on power were evaluated for the mean survival time test and log-rank test. An R package which calculates the power or the sample size for user-specified parameter values and provides power plots for each design parameter is provided. The R package also calculates the probability that the restricted mean survival time is estimable for user-defined trial designs.

RESULTS

Under proportional hazards and late differences in survival, the power of the mean survival time test can approach that of the log-rank test if the restriction time is late. Under early differences, the power of the restricted mean survival time test is higher than that of the log-rank test. Duration of accrual and follow-up have little influence on the power of the restricted mean survival time test. The choice of restriction time, on the other hand, has a large impact on power. Because the power depends on the interplay among the design factors, plotting the relationship between each design parameter and power allows the users to select the designs most appropriate for their trial. When modification is necessary to ensure the difference in restricted mean survival time is estimable, the three available modifications all perform adequately in the scenarios studied.

CONCLUSION

The restricted mean survival time is a survival endpoint that is meaningful to investigators and to patients and at the same time requires less restrictive assumptions. The biggest challenge with this endpoint is selection of the restriction time. We recommend selecting a restriction time that is clinically relevant to the disease and the clinical setting of the trial of interest. The practical considerations and the R package provided in this work are readily available tools that researchers can use to design trials with restricted mean survival time as the primary endpoint.

Cognitive function and neuropathological outcomes: a forward-looking approach

OBJECTIVE

To evaluate the risk of Alzheimer's disease-related neuropathology burden at autopsy given older adults' current cognitive state.

METHOD

Participants included 1,303 individuals who enrolled in the Religious Orders Study (ROS) and 1,789 who enrolled in the Rush Memory and Aging Project (MAP). Cognitive status was evaluated via standardized assessments of global cognition and episodic memory. At the time of analyses, about 50% of participants were deceased with the remaining numbers right censored. Using multi-state Cox proportional hazard models, we compared the cognitive status of all subjects alive at a given age and estimated future risk of dying with different AD-related neuropathologies. Endpoints considered were Braak Stages (0-2, 3-4, 5-6), CERAD (0, 1, 2, 3), and TDP-43 (0, 1, 2, 3) level.

RESULTS

For all three pathological groupings (Braak, CERAD, TDP-43), we found that a cognitive test score one standard deviation below average put individuals at up to three times the risk for being diagnosed with late stage AD at autopsy according to pathological designations. The effect remained significant after adjusting for sex, APOE-e4 status, smoking status, education level, and vascular health scores.

CONCLUSION

Applying multi-state modeling techniques, we were able to identify those at risk of exhibiting specific levels of neuropathology based on current cognitive test performance. This approach presents new and approachable possibilities in clinical settings for diagnosis and treatment development programs.

Differences in genomic abnormalities among African individuals with monoclonal gammopathies using calculated ancestry

Multiple myeloma (MM) is two- to three-fold more common in African Americans (AAs) compared to European Americans (EAs). This striking disparity, one of the highest of any cancer, may be due to underlying genetic predisposition between these groups. There are multiple unique cytogenetic subtypes of MM, and it is likely that the disparity is associated with only certain subtypes. Previous efforts to understand this disparity have relied on self-reported race rather than genetic ancestry, which may result in bias. To mitigate these difficulties, we studied 881 patients with monoclonal gammopathies who had undergone uniform testing to identify primary cytogenetic abnormalities. DNA from bone marrow samples was genotyped on the Precision Medicine Research Array and biogeographical ancestry was quantitatively assessed using the Geographic Population Structure Origins tool. The probability of having one of three specific subtypes, namely t(11;14), t(14;16), or t(14;20) was significantly higher in the 120 individuals with highest African ancestry (≥80%) compared with the 235 individuals with lowest African ancestry (<0.1%) (51% vs. 33%, respectively, p value = 0.008). Using quantitatively measured African ancestry, we demonstrate a major proportion of the racial disparity in MM is driven by disparity in the occurrence of the t(11;14), t(14;16), and t(14;20) types of MM.

Predictors of Durability of Radiological Response in Patients With Small Bowel Crohn's Disease

BACKGROUND

The long-term significance of radiological transmural response (TR) as a treatment goal at the first follow-up scan in small bowel Crohn's disease (CD) has been previously shown. We examined the durability of a long-term strategy of treating to a target of radiological TR and the influence of baseline predictors on the maintenance of TR.

METHODS

Small bowel CD patients between January 1, 2002, and December 31, 2014, were identified with serial computed tomography enterography (CTE)/magnetic resonance enterography (MRE) before and after initiation of therapy or on maintenance therapy. Overall TR (inflammatory lesions with/without strictures) w1as characterized by abdominal radiologists in up to 5 small bowel lesions per patient at each serial scan until last follow-up or small bowel resection, as response, partial response, or nonresponse. The rate of conversion between TR states and transition to surgery, including the effect of baseline patient/disease characteristics, was examined using a multistate model (mstate R-package).

RESULTS

CD patients (n = 150, 705 CTE/MRE) with a median of 4 CTE/MRE during 4.6 years of follow-up, 49% with ileal-only distribution, had 260 examined bowel segments. Conversion from response to partial response/nonresponse was 37.4% per year of follow-up with no transitions seen directly from response to surgery. Current smoking status (hazard ratio [HR], 2.2; 95% confidence interval [CI], 1.1-4.3) and internal penetrating disease at baseline scan (HR, 2.2; 95% CI, 1.2-4.1) were associated with a 2-fold increased risk of transition from partial response/nonresponse to surgery.

CONCLUSIONS

Achievement and maintenance of radiological response is associated with avoidance of small bowel surgery. Continued follow-up with CTE/MRE is recommended to identify loss of response, especially in current smokers and patients with internal penetrating disease at baseline CTE/MRE.

A Topic-modeling Based Framework for Drug-drug Interaction Classification from Biomedical Text

Classification of drug-drug interaction (DDI) from medical literatures is significant in preventing medication-related errors. Most of the existing machine learning approaches are based on supervised learning methods. However, the dynamic nature of drug knowledge, combined with the enormity and rapidly growing of the biomedical literatures make supervised DDI classification methods easily overfit the corpora and may not meet the needs of real-world applications. In this paper, we proposed a relation classification framework based on topic modeling (RelTM) augmented with distant supervision for the task of DDI from biomedical text. The uniqueness of RelTM lies in its two-level sampling from both DDI and drug entities. Through this design, RelTM take both relation features and drug mention features into considerations. An efficient inference algorithm for the model using Gibbs sampling is also proposed. Compared to the previous supervised models, our approach does not require human efforts such as annotation and labeling, which is its advantage in trending big data applications. Meanwhile, the distant supervision combination allows RelTM to incorporate rich existing knowledge resources provided by domain experts. The experimental results on the 2013 DDI challenge corpus reach 48% in F1 score, showing the effectiveness of RelTM.

Serum adipokine and inflammatory markers before and after liver transplantation in recipients with major cardiovascular events

In the nontransplant setting, aberrant serum adipokine levels are associated with cardiovascular (CV) disease. The effects of liver transplantation (LT) on serum adipokine levels and their association with post-LT CV disease have not been studied. A nested case-control study of 77 patients with major CV events more than 4 months after LT analyzed serum adiponectin, resistin, leptin, C-reactive protein, and apolipoprotein levels measured before transplantation and 4, 12, and 24 months after LT. Adiponectin and resistin levels decreased dramatically after LT in all patients. Recipients with CV disease had lower levels of adiponectin and higher levels of resistin, leptin, C-reactive protein, and apolipoprotein B100 than controls. The pre-LT adiponectin level was associated with a 16% increased risk for CV events for every 1 μg/mL decrease in adiponectin [hazard ratio (HR) = 0.84, P = 0.046]. Pre-LT C-reactive protein levels (HR = 1.03, P = 0.047) and 12-month C-reactive protein levels (HR = 1.03, P = 0.03) were associated with CV events after LT. Pre-LT Diabetes (HR = 2.14, P = 0.09), and post-LT resistin (HR = 1.07, P = 0.07), and apolipoprotein B (HR = 1.08, P = 0.08) were associated with a nonsignificantly increased risk of CV events in this small sample size. In conclusion, pre- and post-LT changes in serum adipokine and inflammatory markers may be signals of an increased risk of CV events after LT, but further study is needed.

Abstract 2021: Characterization of myeloma tumors from a multi-ethnic cohort using cytogenetics and genomic analysis

The plasma cell malignancy Multiple Myeloma (MM) is a rare but deadly form of cancer with 5 year survival rates of ∼30%. Epidemiological studies have suggested that MM exacts an especially heavy burden on African American (AA) patients. We examined a multi-ethnic cohort of MM tumors to assess potential differences in the frequency of molecular events in tumors derived from either African American (AA) or European American (EA) patients. The frequency of 14q32 translocation breakpoints at the IgH locus was compared among a series of data from 115 AA MM patients from three studies and EA MM from the Eastern Cooperative Oncology Group (ECOG). In addition, we analyzed matching genome-wide copy number and transcriptional data among 45 AA MM tumors and 196 EA MM tumors to determine differences in the frequency of molecular events in tumors from these populations.

No differences were found for specific translocation subtypes; however, we detected a statistically significant difference in the overall frequency of IgH translocations between the two groups, which occurred more frequently in EA patients (52% vs. 40%; p = 0.032). When assessing genomic copy number events previously shown to be associated with poor outcome in MM, frequencies of alterations were not significant after correcting for multiple testing. Furthermore, there was not a significant difference in the association of high-risk disease using expression profiling. Our study represents the first comprehensive assessment of the frequency and distribution of molecular alterations in MM tumors from both AA and EA patients and could help shed light on possible biological features associated with population differences in incidence and outcome in MM.

Citation Format: Angela S. Baker, Esteban Braggio, Susana Jacobus, Sungwon Jung, Dirk Larson, Terry Therneau, Angela Dispenzieri, Scott A. Van Wier, Gregory Ahmann, Joan Levy, Louise Perkins, Seungchan Kim, Kim Henderson, David Vesole, S. Vincent Rajkumar, Dianne F. Jelinek, John Carpten, Rafael Fonseca. Characterization of myeloma tumors from a multi-ethnic cohort using cytogenetics and genomic analysis. [abstract]. In: Proceedings of the 104th Annual Meeting of the American Association for Cancer Research; 2013 Apr 6-10; Washington, DC. Philadelphia (PA): AACR; Cancer Res 2013;73(8 Suppl):Abstract nr 2021. doi:10.1158/1538-7445.AM2013-2021

Predictors of pretransplant dropout and posttransplant recurrence in patients with perihilar cholangiocarcinoma

We have previously reported excellent outcomes with liver transplantation for selected patients with early-stage perihilar cholangiocarcinoma (CCA) following neoadjuvant chemoradiotherapy. Our aim was to identify predictors of dropout before transplantation and predictors of cancer recurrence after transplantation. We reviewed all patients with unresectable perihilar CCA treated with neoadjuvant chemoradiation in anticipation for transplantation between 1993 and 2010. Predictors were identified by univariate and multivariate Cox regression analysis of clinical variables. In total, 199 patients were enrolled, of whom 62 dropped out and 131 underwent transplantation at our institution, with six undergoing transplantation elsewhere. Predictors of dropout were carbohydrate antigen 19-9 (CA 19-9) ≥ 500 U/mL (hazard ratio [HR] 2.3; P = 0.04), mass ≥ 3 cm (HR 2.1; P = 0.05), malignant brushing or biopsy (HR 3.6; P = 0.001), and Model for End-Stage Liver Disease (MELD) score ≥ 20 (HR 3.5; P = 0.02). Posttransplant, recurrence-free 5-year survival was 68%. Predictors of recurrence were elevated CA 19-9 (HR 1.8; P = 0.01), portal vein encasement (HR 3.3; P = 0.007), and residual tumor on explant (HR 9.8; P < 0.001). Primary sclerosing cholangitis (PSC), age, history of cholecystectomy, and waiting time were not independent predictors.

CONCLUSION

Outcome following neoadjuvant chemoradiation and liver transplantation for perihilar CCA is excellent. Risk of dropout is related to patient and tumor characteristics and this can be used to guide patient counseling before enrollment. Recurrence risk is mostly associated with presence of residual cancer on explant. Patients with PSC do not have an independent survival advantage over de novo patients, but present with more favorable tumor characteristics.

Number of lymph nodes evaluated: prognostic value in pancreatic adenocarcinoma

INTRODUCTION

The impact of the number of lymph node (LN) evaluated pathologically on accurate staging is unknown. Our primary aim was to determine a minimum number of evaluated LN needed to provide accurate staging of pancreatic cancer.

METHODS

Four hundred ninety-nine patients underwent a curative pancreatectomy for pancreatic adenocarcinoma cancer from 1981-2007. The probability of understaging a patient as N0 was estimated based on the number of LN evaluated. The prognostic value of LN ratio (LNR) was assessed.

RESULTS

Survival for node-negative (pN0) patients with <11 LN examined was worse than for pN0 patients with ≥11 LNs with a hazard ratio (95 % CI) of 1.33 (1.1-1.7, p = 0.01) with 3-year survivals of 32 vs. 50%, respectively. Three-year survival for pN1 patients with <11 nodes evaluated was similar to pN1 patients with ≥11 nodes (25 vs. 30%). LNR ≥ 0.17 predicted worse survival with hazard ratio of 1.76 (1.3-2.4, p = 0.001) than LNR < 0.17; 3-year survivals were 37 vs. 19%.

CONCLUSION

Patients with "N0" disease with <11 LN evaluated pathologically have worse survival, suggesting that metastatic nodes were missed by evaluating too few nodes. For pN1 patients, LNR stratifies survival of patient cohorts more accurately. Adequate staging of pancreatic cancer requires pathologic evaluation of ≥11 LNs.

Abstract 5064: Cytogenetic and somatic characterization of myeloma tumors from a multi-ethnic cohort

Previous epidemiological studies have suggested that African Americans (AA) are twice as likely to be diagnosed with and to die from Multiple Myeloma (MM) as compared to European Americans (EA). In recent years, however, mortality rates have become more comparable between these groups. Although socioeconomic factors could play a major role to this potential disparity, biological factors may also be responsible for influencing these differences in either incidence and/or mortality. Here we examined a multi-ethnic cohort of MM tumors to assess potential differences in the frequency of molecular events in tumors derived from either African American (AA) or European American (EA) patients. First, we compared the frequency of 14q32 translocation breakpoints at the IgH locus among a series of data from 115 AA MM patients from three studies and EA MM from the Eastern Cooperative Oncology Group (ECOG). Furthermore, we analyzed matching genome-wide copy number and transcriptional data among 47 AA MM tumors and 196 EA MM tumors to determine differences in the frequency of molecular events in tumors from these populations. Univariate analyses were conducted using the Fisher's Exact Test and p-values were calculated to measure statistical significance. We observed statistically significant differences in the frequency of IgH translocations and somatic copy number alterations. Although we did not detect differences for specific translocation subtypes, we saw a statistically significant difference in the overall frequency of IgH translocations between the two groups, which occurred more frequently in EA patients (p = 0.032). When assessing genomic copy number events previously shown to be associated with poor outcome in MM, chromosome 13 monosomy was more frequent in hyperdiploid MM from AA patients and significant after correcting for multiple testing using the Bonferoni correction method (p = 0.001). Moreover, when using the Benjamini-Hochberg multiple testing correction method, chromosome 13 monosomy remained significantly different among hyperdiploid AA patients (p = 0.0121). Chromosome 1q amplification was also significant among hyperdiploid tumors from EA patients when using this correction method (p = 0.036). No differences in the frequency of hyperdiploid status or gene expression-based subtypes were detected when comparing tumors from AA and EA patients. Our study represents the first comprehensive assessment of the frequency and distribution of molecular alterations in MM tumors from both AA and EA patients and could help shed light on possible biological features associated with population differences in incidence and outcome in MM.

Citation Format: {Authors}. {Abstract title} [abstract]. In: Proceedings of the 103rd Annual Meeting of the American Association for Cancer Research; 2012 Mar 31-Apr 4; Chicago, IL. Philadelphia (PA): AACR; Cancer Res 2012;72(8 Suppl):Abstract nr 5064. doi:1538-7445.AM2012-5064

Estimating underreported N2 disease in rectal cancer patients with low lymph node counts

BACKGROUND

The variability in the number of lymph nodes examined needs to be taken into account for adequate staging. The definition of nodal staging was refined by quantifying the likelihood of N2 disease when the patient had fewer than four positive LN.

METHODS

In a retrospective study a total of 548 patients with node positive rectal cancer and curative surgery between 1990 and 2006 were identified. The misclassification of pN staging was estimated with a Bayesian computation. The prognostic value of the calculated probability, lymph node ratio (LNR), and nodal stage was assessed with Cox proportional hazard regression.

RESULTS

A probability of understaging of 40% or more indicated worse prognosis of cancer-specific survival (CSS) with hazard ratio 2.6 (95%CI: 1.8-3.9, P < 0.001). The concordance index of a multivariate model with probability of N2 disease as a prognostic factor for survival was 0.68 for all patients and 0.75 for patients with less than 10 lymph nodes examined.

CONCLUSION

Utilizing estimated probabilities of N2 disease improves our ability to predict survival, in particular in patients with low LN count. These probabilities allow for a simple rule in patient counseling and clinical decision making.

Design and rationale of the HCC BRIDGE study in China: a longitudinal, multicenter cohort trial in hepatocellular carcinoma

BACKGROUND

More than 50% of the worldwide cases of hepatocellular carcinoma occur in China, and this malignancy currently represents the country's second leading cause of cancer death in cities and the leading cause in rural areas. Despite recent advances in the control and management of hepatocellular carcinoma within China, this disease remains a major health care issue. The global HCC BRIDGE study, designed to assess patterns of hepatocellular carcinoma therapy use and associated outcomes across real-world clinical practice, has recently been expanded as a national study in China, allowing a detailed analysis of hepatocellular carcinoma in this important country.

METHODS/DESIGN

The global HCC BRIDGE study is a multiregional longitudinal cohort trial including patients newly diagnosed with hepatocellular carcinoma between January 1, 2005, and June 30, 2011, who are receiving treatment for hepatocellular carcinoma via sites in the Asia-Pacific, European, and North American regions. The HCC BRIDGE China national study comprises the portion of the global HCC BRIDGE study conducted within mainland China. Patients will be followed from time of diagnosis of hepatocellular carcinoma (post-January 1, 2005) to time of death or December 31, 2011, whichever comes first. Data will be collected on demographic/clinical characteristics, relevant laboratory values, hepatocellular carcinoma/underlying liver disease treatment, tumor response, adverse events, hospitalizations, and overall survival. The primary study end point is overall survival; secondary end points are disease progression, treatment-limiting adverse events, and treatment failure.

RESULTS

At the time of writing, 15 sites have selected for participation across all 7 traditional regions of China (North, North-East, East, South, South-West, North-West, and Central). The anticipated study population from the China national study is approximately 9000 patients.

DISCUSSION

Findings from the HCC BRIDGE China national study, the first geographically representative study of hepatocellular carcinoma in China, will contribute to the understanding of patterns of therapy use and related clinical outcomes and will provide further information on continuing unmet needs for hepatocellular carcinoma throughout this important country.

Nonalcoholic fatty liver disease increases risk of death among patients with diabetes: a community-based cohort study

OBJECTIVES

The significance of nonalcoholic fatty liver disease (NAFLD) among patients with diabetes is unknown. We sought to determine whether a diagnosis of NAFLD influenced mortality among a community-based cohort of patients with type II diabetes mellitus.

METHODS

A total of 337 residents of Olmsted County, Minnesota with diabetes mellitus diagnosed between 1980 and 2000 were identified using the Rochester Epidemiology Project and the Mayo Laboratory Information System, and followed for 10.9 + or - 5.2 years (range 0.1-25). Survival was analyzed using Cox proportional hazards modeling, with NAFLD treated as a time-dependent covariate.

RESULTS

Among the 337 residents, 116 were diagnosed with NAFLD 0.9 + or - 4.6 years after diabetes diagnosis. Patients with NAFLD were younger, and more likely to be female and obese. Overall, 99/337 (29%) patients died. In multivariate analysis to adjust for confounders, overall mortality was significantly associated with a diagnosis of NAFLD (hazard ratio (HR) 2.2; 95% confidence interval (CI) 1.1, 4.2; P=0.03), presence of ischemic heart disease (HR 2.3; 95% CI 1.2, 4.4), and duration of diabetes (HR per 1 year, 1.1; 95% CI 1.03, 1.2). The most common causes of death in the NAFLD cohort were malignancy (33% of deaths), liver-related complications (19% of deaths), and ischemic heart disease (19% of deaths). In adjusted multivariate models, NAFLD was borderline associated with an increased risk of dying from malignancy (HR 2.3; 95% CI 0.9, 5.9; P=0.09) and not from cardiovascular disease (HR 0.9; 95% CI 0.3, 2.4; P=0.81).

CONCLUSIONS

The diagnosis of NAFLD is associated with an increased risk of overall death among patients with diabetes mellitus.

IgM monoclonal gammopathy of undetermined significance and smoldering Waldenström's macroglobulinemia

Immunoglobulin M monoclonal gammopathy of undetermined significance (IgM-MUS) was diagnosed in 213 Mayo Clinic patients who were residents of 11 counties in southeastern Minnesota from 1960 to 1994. During long-term follow-up, 29 (14%) developed non-Hodgkin lymphoma (n = 17), Waldenström's macroglobulinemia (WM; n = 6), chronic lymphocytic leukemia (n = 3), and AL amyloidosis (n = 3) with relative risks of 15-, 262-, 6-, and 16-fold, respectively. The cumulative probability of progression to one of these disorders was 10% at 5 years, 18% at 10 years, and 24% at 15 years, approximately 1.5% per year. Smoldering WM was identified in 48 patients at Mayo Clinic from 1974 to 1995. During follow-up, 33 of the 48 patients progressed to symptomatic WM. The median time to progression was 4.6 years. The risk of progression to WM was 6% at 1 year, 39% at 3 years, and 55% at 5 years.

The application of real-time PCR to the analysis of T cell repertoires

The diversity of T-cell populations is determined by the spectrum of antigen-specific T-cell receptors (TCRs) that are heterodimers of alpha and beta subunits encoded by rearranged combinations of variable (AV and BV), joining (AJ and BJ), and constant region genes (AC and BC). We have developed a novel approach for analysis of beta transcript diversity in mice with a real-time PCR-based method that uses a matrix of BV- and BJ-specific primers to amplify 240 distinct BV-BJ combinations. Defined endpoints (Ct values) and dissociation curves are generated for each BV-BJ combination and the Ct values are consolidated in a matrix that characterizes the beta transcript diversity of each RNA sample. Relative diversities of BV-BJ combinations in individual RNA samples are further described by estimates of scaled entropy. A skin allograft system was used to demonstrate that dissection of repertoires into 240 BV-BJ combinations increases efficiency of identifying and sequencing beta transcripts that are overrepresented at inflammatory sites. These BV-BJ matrices should generate greater investigation in laboratory and clinical settings due to increased throughput, resolution and identification of overrepresented TCR transcripts.

Mortality attributable to cholestatic liver disease in the United States

In the past 2 decades, important advances have been made in the treatment of cholestatic liver diseases, including primary biliary cirrhosis (PBC) and primary sclerosing cholangitis (PSC). Whether these new therapies have had demonstrable impact on mortality on a population-wide scale has not been evaluated. This study describes the age-specific and sex-specific mortality rates from PBC and PSC in the United States between 1980 and 1998, based on the Multiple Cause of Death files. Age-specific and sex- specific mortality rates from PBC and PSC were calculated. The multivariable Poisson model was used to evaluate temporal changes in mortality rates. In 1998, the total age-adjusted and sex-adjusted PBC-related mortality rate was 0.24 per 100,000, and the age-adjusted and sex-adjusted PSC-related mortality rate was 0.23 per 100,000. During the observation period, PBC-related mortality significantly decreased over time in women younger than 65 years, and in men of all age groups, whereas in older women this number increased over time. PSC-related mortality remained essentially stable, except in men 65 years of age or older.

CONCLUSION

Since the early 1980s, significant changes in mortality from PBC have occurred. The most noticeable change was an increase in the age of death, which indicates prolongation of survival. These changes may be attributable to liver transplantation or ursodeoxycholic acid. In contrast, mortality from PSC remained largely unchanged, highlighting the need for more effective therapeutic strategies.

Evidence-based incorporation of serum sodium concentration into MELD

BACKGROUND & AIMS

Serum sodium (Na) concentrations have been suggested as a useful predictor of mortality in patients with end-stage liver disease awaiting liver transplantation.

METHODS

We evaluated methods to incorporate Na into model for end-stage liver disease (MELD), using a prospective, multicenter database specifically created for validation and refinement of MELD. Adult, primary liver transplant candidates with end-stage liver disease were enrolled.

RESULTS

Complete data were available in 753 patients, in whom the median MELD score was 10.8 and sodium was 137 mEq/L. Low Na (<130 mEq/L) was present in 8% of patients, of whom 90% had ascites. During the study period, 67 patients (9%) died, 243 (32%) underwent transplantation, 73 (10%) were withdrawn, and 370 were still waiting. MELD score and Na, at listing, were significant (both, P < .01) predictors of death within 6 months. After adjustment for MELD score and center, there was a linear increase in the risk of death as Na decreased between 135 and 120 mEq/L. A new score to incorporate Na into MELD was developed: "MELD-Na" = MELD + 1.59 (135 - Na) with maximum and minimum Na of 135 and 120 mEq/L, respectively. In this cohort, "MELD-Na" scores of 20, 30, and 40 were associated with 6%, 16%, and 37% of risk of death within 6 months of listing, respectively. If this new score were used to allocate grafts, it would affect 27% of the transplant recipients.

CONCLUSIONS

We demonstrate an evidence-based method to incorporate Na into MELD, which provides more accurate survival prediction than MELD alone.

Relationship of patient survival and chromosome anomalies detected in metaphase and/or interphase cells at diagnosis of myeloma

The clinical efficacy of evaluating genetic anomalies in metaphase cells versus interphase nuclei for multiple myeloma (MM) is poorly understood. Therefore, survival for 154 patients with newly diagnosed untreated MM was compared with results from analysis of metaphase and interphase cells. Metaphases were studied by conventional cytogenetics and fluorescent-labeled DNA probes (fluorescence in situ hybridization [FISH]), whereas inter-phase nuclei were evaluated only by FISH. All FISH studies were done using DNA probes to detect t(4;14)(p16;q32), t(11;14)(q13;q32), t(14;16)(q32;q23), del(17) (p13.1), and chromosome 13 anomalies. Metaphases were abnormal by cytogenetics and/or metaphase FISH in 61 (40%) patients. Abnormal interphase nuclei were observed in 133 (86%) patients, including each patient with abnormal metaphases. FISH was a necessary adjunct to cytogenetics to detect t(4;14) and t(14;16) in metaphase cells. Patient survival was especially poor for patients with greater than 50% abnormal interphase nuclei, although this result was more likely due to level of plasma cells than specific chromosome anomalies. For metaphase data, patients with t(4;14), t(14;16), del(17) (p13.1), and/or chromosome 13 anomalies (primarily monosomy 13) had poor survival. A different outcome was observed for interphase data as patients with t(4;14) or t(14;16) had poor survival, whereas patients with chromosome 13 anomalies had intermediate survival: interphase FISH did not substitute for metaphase analysis.

Sharpening spots: correcting for bleedover in cDNA array images

For cDNA array methods that depend on imaging of a radiolabel, we show that bleedover of one spot onto another, due to the gap between the array and the imaging media, can be a major problem. The images can be sharpened, however, using a blind convolution method based on the EM algorithm. The sharpened images look like a set of donuts, which concurs with our knowledge of the spotting process. Oversharpened images are actually useful as well, in locating the centers of each spot.

Impact of nutritional status on outcomes after liver transplantation

BACKGROUND

Poor preoperative nutritional status has been reported to be associated with adverse outcomes after liver transplantation. Published data are, however, conflicting, with methods of preoperative nutritional assessment and postoperative outcomes varying between studies.

METHODS

We prospectively studied the predictive value of preoperative nutritional status for adverse outcomes after liver transplantation. Assessment of preoperative nutritional status included: body cell mass determination, subjective global assessment, anthropometry, handgrip dynamometry, biochemical and amino acid profile, Child's score, and dual-energy x-ray absorptiometry. Death, intensive care unit (ICU) length of stay > or =4 days, hospital length of stay > or =15 days, blood usage > or =36 U of blood products, infection, rejection, and global resource utilization (an index of cost) greater than the median were considered poor outcomes.

RESULTS

Fifty-three patients were studied. Longer ICU stay was associated with lower handgrip strength (P<0.01) and lower aromatic amino acid levels (P<0.01). Longer total hospital stay and the development of infections were associated with lower branched chain amino acid levels (P<0.01 and <0.001, respectively). Acute cellular rejection was associated with lower total body fat (P<0.001) and higher triglyceride levels (P<0.02). Neither death nor higher global resource utilization was associated with any preoperative nutritional parameter.

CONCLUSIONS

Lower preoperative handgrip strength and branched chain amino acid levels are associated with longer ICU stays and increased likelihood of posttransplant infections. In our program, in which nutritional support was provided to potential recipients exhibiting malnourishment, none of the measured nutritional parameters were associated with mortality or greater global resource utilization.

Impact of nutritional status on outcomes after liver transplantation

BACKGROUND

Poor preoperative nutritional status has been reported to be associated with adverse outcomes after liver transplantation. Published data are, however, conflicting, with methods of preoperative nutritional assessment and postoperative outcomes varying between studies.

METHODS

We prospectively studied the predictive value of preoperative nutritional status for adverse outcomes after liver transplantation. Assessment of preoperative nutritional status included: body cell mass determination, subjective global assessment, anthropometry, handgrip dynamometry, biochemical and amino acid profile, Child's score, and dual-energy x-ray absorptiometry. Death, intensive care unit (ICU) length of stay > or =4 days, hospital length of stay > or =15 days, blood usage > or =36 U of blood products, infection, rejection, and global resource utilization (an index of cost) greater than the median were considered poor outcomes.

RESULTS

Fifty-three patients were studied. Longer ICU stay was associated with lower handgrip strength (P<0.01) and lower aromatic amino acid levels (P<0.01). Longer total hospital stay and the development of infections were associated with lower branched chain amino acid levels (P<0.01 and <0.001, respectively). Acute cellular rejection was associated with lower total body fat (P<0.001) and higher triglyceride levels (P<0.02). Neither death nor higher global resource utilization was associated with any preoperative nutritional parameter.

CONCLUSIONS

Lower preoperative handgrip strength and branched chain amino acid levels are associated with longer ICU stays and increased likelihood of posttransplant infections. In our program, in which nutritional support was provided to potential recipients exhibiting malnourishment, none of the measured nutritional parameters were associated with mortality or greater global resource utilization.

Hepatopulmonary syndrome: a prospective study of relationships between severity of liver disease, PaO(2) response to 100% oxygen, and brain uptake after (99m)Tc MAA lung scanning

BACKGROUND

Because of the spectrum of intrapulmonary vascular dilation that characterizes hepatopulmonary syndrome (HPS), PaO(2) while breathing 100% oxygen varies. Abnormal extrapulmonary uptake of (99m)Tc macroaggregated albumin (MAA) after lung perfusion is common.

GOAL

To describe relationships between (1) severity of liver disease measured by the Child-Pugh (CP) classification; (2) PaO(2) while breathing room air (RA) and 100% oxygen on 100% oxygen; and (3) extrapulmonary (brain) uptake of (99m)Tc MAA after lung scanning.

METHODS AND PATIENTS

We prospectively measured PaO(2) on RA, PaO(2) on 100% oxygen, and brain uptake after lung perfusion of (99m)Tc MAA in 25 consecutive HPS patients.

RESULTS

Mean PaO(2) on RA, PaO(2) on 100% oxygen, PaCO(2) on RA, and (99m)Tc MAA brain uptake were similar when categorized by CP classification. Brain uptake was abnormal (> or = 6%) in 24 patients (96%). Brain uptake was 29 +/- 20% (mean +/- SD) and correlated inversely with PaO(2) on RA (r = -0.57; p<0.05) and PaO(2) on 100% oxygen (r = -0.41; p<0.05). Seven patients (28%) had additional nonvascular pulmonary abnormalities and lower PaO(2) on 100% oxygen (215+/-133 mm Hg vs 391+/-137 mm Hg; p<0.007). Eight patients (32%) died. Mortality in patients without coexistent pulmonary abnormalities was associated with greater brain uptake of (99m)Tc MAA (48+/-18% vs 25+/-20%; p<0.04) and lower PaO(2) on RA (40+/-7 mm Hg vs 57+/-11 mm Hg; p<0.001).

CONCLUSION

The degree of hypoxemia associated with HPS was not related to the CP severity of liver disease. HPS patients with additional nonvascular pulmonary abnormalities exhibited lower PaO(2) on 100% oxygen. Mortality was associated with lower PaO(2) on RA, and with greater brain uptake of (99m)Tc MAA.

A prospective study of CD38/45 flow cytometry and immunofluorescence microscopy to detect blood plasma cells in patients with plasma cell proliferative disorders

Malignant plasma cells can be detected in the blood of patients with multiple myeloma (MM) using flow cytometry (FC), immunofluorescence microscopy (IM), or a variety of molecular techniques. Increased numbers of light chain-restricted blood plasma cells as detected by IM is associated with a diagnosis of overt MM and a decreased overall survival. The IM technique is time consuming; therefore, a prospective study was designed to test whether CD38 CD45 FC could simplify the procedure. Blood samples from 769 patients with plasma cell proliferative disorders were studied prospectively by FC and IM over a one-year period. The FC technique was performed on 1 ml of whole blood after ammonium chloride red blood cell lysis and utilized anti-CD38PE and anti-CD45PerCP. The number of CD38+ 45- events were enumerated and compared to the number of light chain-restricted plasma cells detected by the standard IM technique. In 46% (353/769) of cases > or = 1 CD38+ CD45- events were detected by FC whereas IM was positive for light chain restricted plasma cells in 33%; there was concordance between FC and IM in 73% of cases. In 20% of cases FC was positive and IM was negative; however, in 7% of cases FC was negative yet light chain-restricted plasma cells were detected by IM. FC was positive in 88% (134/153) of cases where the IM technique showed a high number of circulating plasma cells. This study demonstrates that two-color CD38/45 FC identifies most cases with a high IM result and reduces the workload in the clinical laboratory. The prognostic implications of a positive FC screen but a negative IM will require long-term patient follow-up.

Abnormal cytogenetics predict poor survival after high-dose therapy and autologous blood cell transplantation in multiple myeloma

We compared the prognostic value of conventional cytogenetic analysis and established factors such as beta2-microglobulin and plasma cell labeling index in 70 patients undergoing autologous blood cell transplantation for multiple myeloma. Patients underwent transplantation 5 to 88 months (median, 20 months) after the initial diagnosis of myeloma. Factors studied were age, sex, beta2-microglobulin, response to prior therapy, plasma cell labeling index, cytogenetic analysis, bone marrow plasma cell percentage, lactate dehydrogenase and C-reactive protein. Twenty-eight of 65 patients (43%) had abnormal marrow cytogenetics. Overall survival measured from transplantation was significantly better in patients with normal cytogenetics than in those with abnormal cytogenetics (median survival, 25 vs 12 months, P = 0.003). Progression-free survival was better, with median times of 12 vs7 months, respectively (P = 0.005); overall survival measured from the time myeloma was first diagnosed was also longer, with median survivals of 62 and 39 months, respectively (P = 0.001). Median plasma cell labeling index was 1.5% in patients with abnormal cytogenetics and 0. 2% in those with normal cytogenetics (P < 0.001). Abnormal bone marrow cytogenetics predict poor survival after blood cell transplantation for myeloma. There is a significant correlation between abnormal cytogenetics and high plasma cell labeling index, suggesting that certain cytogenetic abnormalities may offer a proliferative advantage to myeloma cells.

Measurement of apoptosis and proliferation of bone marrow plasma cells in patients with plasma cell proliferative disorders

The proliferative rate of malignant plasma cells, as measured by the plasma cell labelling index (PCLI), is an important prognostic factor in multiple myeloma (MM); however, the PCLI alone is probably Inadequate to describe tumour growth because it ignores the idea that myeloma cells may have a reduced rate of apoptosis. The aims of this study were to develop a flow cytometric method to measure the apoptosis index of fresh marrow plasma cells and develop a plasma cell growth index (PCGI) that related both proliferation and apoptosis to disease activity. Marrow aspirates were obtained from 91 patients with plasma cell disorders and the plasma cells in apoptosis were identified by either 7-amino actinomycin-D (7-AAD) or annexin V-FITC three-colour flow cytometry. The median plasma cell apoptotic index (PCAI) for patients with monoclonal gammopathy of undetermined significance (MGUS), smouldering or indolent myeloma (SMM/IMM), and new multiple myeloma (MM) was 5.2, 3.4 and 2.4, respectively (P=0.03, MGUS v MM). The median PCLI for these same patient groups was 0.0, 0.2 and 0.6, respectively (P<0.001, MGUS v MM). The paired PCLI and PCAI for each sample were used to derive the PCGI=2 + [PCLI-(O.1)(PCAI)]. The median PCGI for patients with inactive disease (MGUS, SMM/IMM or amyloidosis) was 1.8 compared to 2.4 for those with active disease (new or relapsed MM) (P<0.001). These results suggest that a decrease in the PCAI may be a factor in the progression from MGUS to SMM to overt MM.

Cost-effectiveness of ursodeoxycholic acid therapy in primary biliary cirrhosis

Ursodeoxycholic acid (UDCA) is a safe and effective treatment for patients with primary biliary cirrhosis (PBC), but the cost of this drug has raised concerns regarding cost-effectiveness. The aim of our study was to determine the cost-effectiveness of UDCA in PBC. We compared the costs and outcomes of managing PBC patients with and without UDCA. From two previously published trials, the effectiveness of UDCA was determined by comparing the annual reduction in the development of ascites, varices, variceal bleeding, encephalopathy, liver transplantation, and death between the treatment groups. Average annual costs for each of these events were estimated based on literature and institutional data. Approximately twice as many major events occurred in the placebo group compared with the UDCA group. The relative risk (RR) of liver transplantation (1.95; 95% CI: 1.14-3.68) and development of esophageal varices (3. 11; 95% CI: 1.57-10.65) were significantly higher in the placebo group compared with the UDCA group. There were no significant increases in the RR of ascites, variceal bleeding, encephalopathy, or death between the two groups. Based on the estimated annual cost of managing these events and the annual costs of UDCA ($2,500), there was an annual cost savings per patient of $1,372. Compared with the placebo group, patients receiving UDCA had a lower incidence of major complications and lower medical care costs.

Syndecan-1 expression on malignant cells from the blood and marrow of patients with plasma cell proliferative disorders and B-cell chronic lymphocytic leukemia

Syndecan-1 is a low-affinity receptor for basic fibroblast growth factor (bFGF). In this study, we used flow cytometry to examine expression of syndecan-1 on monoclonal cells from the blood (n = 37) and marrow (n = 81) of patients with plasma cell (PC) proliferative disorders (PCPD) and blood cells from patients (n = 39) with B cell chronic lymphocytic leukemia (B-CLL). The marrow CD38+CD45- and CD38+CD45+ PC were syndecan-1 positive in all patients with PCPD and there was no difference between patients with monoclonal gammopathy of undetermined significance (MGUS) vs multiple myeloma or cases with vs without bone lesions. In 38% of cases, syndecan-1 expression on the PC was heterogeneous with > or =25% of PC syndecan-1 negative. We found similar syndecan-1 expression on blood and marrow PC in the 36 cases with paired samples. CLL cells were syndecan-1 negative in 97% (38/39) of the cases. Syndecan-1 is a useful marker to detect malignant plasma cells in the blood or marrow; however, it is not helpful in distinguishing MGUS from active myeloma. In addition, syndecan-1 is present on the less mature (CD45+) PC, and there is heterogeneity of expression within and between patients. The relevance of the bFGF bound to myeloma cells via syndecan-1 remains to be elucidated.

TT-virus infection in North American blood donors, patients with fulminant hepatic failure, and cryptogenic cirrhosis

A novel DNA virus, TT-virus (TTV), has been reported in patients with non-A-G posttransfusion hepatitis in Japan. We sought to determine whether TTV infection occurs in North American blood donors and to further determine the prevalence of TTV infection in several groups of patients with liver disease, including patients with cryptogenic cirrhosis and idiopathic fulminant hepatic failure. TTV infection was sought by detection of TTV DNA in serum by polymerase chain reaction (PCR) using primers generated from a conserved region of the TTV genome. Blood donors, patients with cryptogenic cirrhosis, idiopathic fulminant hepatic failure, and patients with other forms of advanced liver disease with and without a history of parenteral exposures were studied. TTV infection was present in 1% (1 of 100) of blood donors, 15% (5 of 33) of patients with cryptogenic cirrhosis, 27% (3 of 11) of patients with idiopathic fulminant hepatic failure, 18% (2 of 11) of patients with a history of exposure to blood products, and 4% (1 of 25) of patients without parenteral risk factors. For all patients tested, a history of prior exposure to blood products was associated with an increased risk of TTV infection (relative risk, 4.5; 90% confidence intervals, 0.6-43.9). We conclude that TTV infection is present among North American blood donors and is common in patients with liver disease, including cryptogenic cirrhosis and fulminant hepatic failure. Further studies are required to determine the role of TTV in the pathogenicity of acute and/or chronic liver disease.

Cost-effectiveness of ultrasound-guided liver biopsy

The risk of a major complication from "blind" percutaneous liver biopsy is reported to be in the range of 0.24% to 3.8%. In a recent randomized trial, patients whose liver biopsies were performed with ultrasonography had a significant reduction in complications requiring hospitalization compared with patients without ultrasound-guided biopsies (0.5% vs. 2.2%, P < .05). Despite this, routine use of ultrasonography for liver biopsies has not been implemented because of controversies with respect to cost-effectiveness. The aim of our study was to analyze the relative cost-effectiveness of performing ultrasound-guided liver biopsies using decision analysis. A decision tree was constructed to compare a strategy of liver biopsy using ultrasonography with a strategy without ultrasonography. The major outcomes included were minor complications such as pain requiring analgesics and major complications, which require hospitalization. Costs included were direct medical costs from the payer's perspective. In our baseline model, the cost from complications per patient with and without ultrasonography was $62 and $129, respectively. The marginal effectiveness expressed as the number of major complications avoided was 1.2/100 liver biopsies. The incremental cost to avoid one major complication was $2,731. The model was most sensitive to the frequency of major complications and the additional cost of ultrasonography. Our decision analysis model suggests that ultrasound-guided liver biopsy is cost-effective. Future studies assessing the efficacy of image-guided liver biopsies should be conducted.

DNA content flow cytometry as a prognostic factor for node-positive breast cancer. The role of multiparameter ploidy analysis and specimen sonication

The DNA content was analyzed in paraffin-embedded material from 167 patients with node-positive breast cancer to learn whether specimen sonication and multiparameter ploidy analysis (MPPA) (using DNA content and light scatter) could improve the strength of ploidy as a prognostic variable. Sonicated specimens were found to have fewer aggregates, a lower percentage of cells in S-phase (%S) and G2M phase than the corresponding nonsonicated specimens. The results using MPPA predicted the prognosis better because they allowed detection of small aneuploid peaks in histograms classified as diploid or tetraploid using DNA content alone. Ploidy was a significant univariate factor, and patients with tetraploid tumors had the best survival. In the multivariate analysis, if other routine factors were examined preferentially, ploidy and %S did not provide additional prognostic information for survival. This study of paraffin-embedded breast cancers suggested that sonication and MPPA may improve the ploidy analysis in certain cases and that tetraploidy may be a favorable ploidy pattern in this group.

A phase II randomized trial of megestrol acetate or dexamethasone in the treatment of hormonally refractory advanced carcinoma of the prostate

The results of a randomized, multicenter, cooperative group trial evaluating hormonal therapy with either megestrol acetate or dexamethasone in advanced, hormonally refractory prostate cancer are reported. Three of 29 patients (approximately 10%) on the megestrol acetate arm experienced an objective response lasting 41, 84, and 202 days, respectively, whereas two of 29 patients (approximately 7%) on the dexamethasone arm achieved an objective response lasting 359 and 512 days, respectively. Twenty of 29 patients (approximately 69%) on the megestrol acetate arm had stable disease lasting for a median duration of 117 days, whereas 21 of 29 patients (72%) on the dexamethasone arm had stable disease for a median duration of 86 days. Median survival of all patients was 9 months from initiation of treatment. The median survival of all patients on the megestrol acetate arm was 268 days compared to 246 days for patients on the dexamethasone arm (P = 0.2). Neither dexamethasone nor megestrol acetate would seem to be of substantive value in altering the progression of advanced, hormonally refractory prostate cancer.

Pattern of failure after radical retropubic prostatectomy for clinically and pathologically localized adenocarcinoma of the prostate: influence of tumor deoxyribonucleic acid ploidy

From 1966 to 1980, 315 patients underwent bilateral pelvic lymphadenectomy and radical retropubic prostatectomy without adjuvant treatment for clinically and pathologically localized adenocarcinoma of the prostate. Followup was 5 to 21 years, with a median of 8 years. The disease was pathological stage A in 24 patients (8%) and pathological stage B in 291 (92%). A total of 45 patients (14.2%) experienced progression. Over-all, 28 patients (8.9%) suffered local recurrence at a mean of 6.6 years postoperatively (median 5.5 years). Local recurrence was noted as late as 15 years postoperatively. Over-all, systemic progression was observed in 25 patients (8%) after a mean of 4.7 years (median 6 years). Eight patients (2.5%) experienced local and systemic failure. The projected local and systemic failure rates at 15 years were 22% and 15%, respectively. Disease-specific survival at 15 years was 93%, since only 11 patients (3.4%) died of prostate cancer. In an age-matched case control analysis, after all prognostic variables were analyzed (Mayo grade, Gleason score, capsule involvement, number of foci, volume of tumor and deoxyribonucleic acid tumor ploidy), progression was related to nondiploid deoxyribonucleic acid tumor ploidy (p less than 0.0004) as determined by flow cytometry in 63% of the patients who evidenced progression versus 8% of the nonrecurrent group.

The mortality risk associated with digitalis treatment after myocardial infarction

We examined the effects of digitalis therapy on postinfarction mortality throughout a 24-month to 48-month follow-up in 867 patients who survived an acute myocardial infarction. During follow-up, 145 patients died (16.7% mortality). At the time of hospital discharge, 31% of the patients were taking digitalis. The digitalis-treated patients were older, had more medical-cardiac risk factors, and had a higher mortality rate throughout the follow-up than the nondigitalis-treated patients. Statistical techniques were used to adjust for clinical imbalances between the digitalis-treated patients and nondigitalis-treated patients. The survival analysis (n = 728 patients) utilized the Cox regression model, and the digitalis-associated mortality risk was identified only after all significant covariates were allowed, so that mortality could be predicted as accurately as possible. Digitalis therapy was associated with a significantly increased postinfarction mortality risk after adjustment for the predictor covariates (relative risk 2.3, 95% confidence interval 1.4-3.7, p less than 0.001). The findings from this large multicenter study suggest that it would be prudent to exercise caution in the use of digitalis in postinfarction patients.

A two-stage design for randomized trials with binary outcomes

A two-stage rule for a randomized clinical trial with a dichotomous outcome is presented. It is based on a rule discussed by Ellenberg and Eisenberger (Cancer Treat Rep 69:1147-1154, 1985). The rule is shown to lead to a reduction in expected sample size if the test treatment is ineffective, with almost no loss in power. All the mathematical formulas required to design such a rule are given, as well as some time-saving approximations. The accuracy of the approximations is shown.