Targeted learning ensembles for optimal individualized treatment rules with time-to-event outcomes.

We consider estimation of an optimal individualized treatment rule when a high-dimensional vector of baseline variables is available. Our optimality criterion is with respect to delaying the expected time to occurrence of an event of interest. We use semiparametric efficiency theory to construct estimators with properties such as double robustness. We propose two estimators of the optimal rule, which arise from considering two loss functions aimed at directly estimating the conditional treatment effect and recasting the problem in terms of weighted classification using the 0-1 loss function. Our estimated rules are ensembles that minimize the crossvalidated risk of a linear combination in a user-supplied library of candidate estimators. We prove oracle inequalities bounding the finite-sample excess risk of the estimator. The bounds depend on the excess risk of the oracle selector and a doubly robust term related to estimation of the nuisance parameters. We discuss the convergence rates of our estimator to the oracle selector, and illustrate our methods by analysis of a phase III randomized study testing the efficacy of a new therapy for the treatment of breast cancer.

Management of regional lymph nodes in the elderly melanoma patient: patient selection, accuracy and prognostic implications.

BACKGROUND: Among older melanoma patients, lymphatic mapping failure, lower rates of SLN positivity and poor prognosis are reported reasons for omission of sentinel lymph node biopsy (SLNB). We investigated reasons for non-compliance with guidelines, sensitivity and prognostic significance of SLNB and completion lymphadenectomy (CLND) for elderly melanoma patients. METHODS: Retrospective review of patients ≥65 years with ≥1 mm thick melanoma treated at a single Institution. Wilcoxon, chi-square and Fisher's exact tests were used for analysis as appropriate. Univariable and multivariable Cox regressions were used to analyze time-to-event variables. RESULTS: 72 of 358 patients (20%) did not undergo SLNB. Reasons for omission included selective neck dissection in 26 (7%), patient refusal in 11 (3%), physician recommendation in 15 (4%) and significant comorbidities in 8 (2%). Of the 286 patients undergoing SLNB, only 5 (1.7%) had lymphatic mapping failures. 76 patients (26.6%) were SLN-positive. The sensitivity of SLNB was 90.5%, the negative predictive value was 96.3% and the false negative rate was 3.8%. Sixty-seven (88%) SLN-positive patients underwent CLND and 10 (15%) had positive non-SLNs. Reasons for omission of CLND included patient refusal in 3 (4%), surgeon recommendation in 5 (7%) and postoperative complication in 1 (1%). SLN and non-SLN status were independently associated with disease-free survival. SLN status was independently associated with melanoma-specific survival. CONCLUSIONS: SLNB was successful in 98.7% of elderly patients with high sensitivity and a low false negative rate. Only 2% of our elderly patients appeared too frail for SLNB. Age alone should not be a contraindication to SLNB and CLND for melanoma.

EGFR expression is associated with decreased benefit from trastuzumab in the NCCTG N9831 (Alliance) trial.

BACKGROUND: Epidermal growth factor receptor (EGFR) has been hypothesised to modulate the effectiveness of anti-HER2 therapy. We used a standardised, quantitative immunofluorescence assay and a novel EGFR antibody to evaluate the correlation between EGFR expression and clinical outcome in the North Central Cancer Treatment Group (NCCTG) N9831 trial. METHODS: Tissue microarrays were constructed that allowed analysis of 1365 patients randomly assigned to receive chemotherapy alone (Arm A), sequential trastuzumab after chemotherapy (Arm B) and chemotherapy with concurrent trastuzumab (Arm C). Measurement of EGFR was performed using the EGFR antibody, D38B1, on the fluorescence-based AQUA platform. The result was validated using an independent retrospective metastatic breast cancer cohort (n=130). RESULTS: Epidermal growth factor receptor assessed as a continuous (logarithmic transformed) variable shows an association with disease-free survival in Arm C (P=0.009) but not in Arm A or B. High EGFR expression was associated with worse outcome (Hazard ratio (HR)=2.15; 95% CI 1.28-3.60, P=0.004). Validation in a Greek metastatic breast cancer cohort showed an HR associated with high EGFR expression of 1.92 (P=0.0073). CONCLUSIONS: High expression of EGFR appears to be associated with decreased benefit from adjuvant concurrent trastuzumab. Since other treatment options exist for HER2-driven tumours, further validation of these data may select patients for alternative or additive therapy.

Breast cancer and aging: results of the U13 conference breast cancer panel.

Breast cancer is predominantly a disease of older women, yet there is a knowledge gap due to the persisting misalignment between the age distribution of women with breast cancer and the age distribution of participants in clinical trials. The purpose of this report is to state the U13 conference breast cancer panel's recommendations regarding therapeutic clinical trials that will fill gaps in knowledge regarding the care of older patients with breast cancer. The U13 conference was a collaboration between the Cancer and Aging Research Group and the National Institute on Aging and the National Cancer Institute (NCI). Clinical trials should be developed for frail and vulnerable patients who would not enroll on the standard phase III trials, as well as efforts need to be made to increase enrollment of fit older patients on standard phase III trials. As a result of this conference, panel members are working with the NCI and cooperative groups to address these knowledge gaps. With the aging population and increasing incidence of breast cancer with age, it is essential to study the feasibility, toxicity, and efficacy of cancer therapy in this at-risk population.

Transformation of low grade glioma and correlation with outcome: an NCCTG database analysis.

Glioblastomas (GBM) may originate de novo (primary), or following transformation from a lower grade glioma (secondary), and it has been postulated that these tumors may have different biological behaviors. We performed a correlative analysis involving 204 patients with glioma treated prospectively on NCCTG clinical trials. Central pathology review of tumor tissues taken at the time of initial diagnosis and at recurrence were performed in all patients. Tumors progressed from low (WHO grade 2) to high (grade 3-4) at recurrence in 45% low grade oligodendroglioma patients, in 70% with low grade oligoastrocytoma, and 74% with low grade astrocytoma (P = 0.031). Median overall survival (OS) from initial diagnosis varied by histology: oligodendroglioma, 8.8 years; (95% CI 5.7-10.2); oligoastrocytoma, 4.4 years (95% CI 3.5-5.6); astrocytoma grade 2 3.1 years (astrocytoma grade 2-4, 2.1 years) (95% CI 1.7-2.5, P < 0.001). Mean time to recurrence (TTR) also varied between patients with de novo GBM, those secondary GBM, and those that remained non-GBM at recurrence (1.1 ± 1.1 vs. 2.9 ± 1.8 vs. 4.0 ± 2.9 years, respectively, P < 0.001). Median OS from time of recurrence also varied between these three categories (0.7 years, 95% CI: 0.5-1.1 vs. 0.6 years, CI: 0.5-1.0 vs. 1.4 years, 95% CI: 1.1-2.0, respectively) (P < 0.001). At time of relapse, transformation to higher grade is frequent in low grade pure and mixed astrocytomas, but is observed in less than half of those with low grade oligodendroglioma. From time of recurrence, OS was not significantly different for those with primary versus secondary GBM, and it may thus be reasonable include patients with secondary GBM in clinical therapeutic trials for recurrent disease.

Do cardiovascular risk factors confer the same risk for cardiovascular outcomes in rheumatoid arthritis patients as in non-rheumatoid arthritis patients?

OBJECTIVE: To compare the frequency of traditional cardiovascular (CV) risk factors in rheumatoid arthritis (RA) compared to non-RA subjects, and examine their impact on the risk of developing selected CV events (myocardial infarction (MI), heart failure (HF) and CV death) in these two groups. METHODS: We examined a population-based incidence cohort of subjects with RA (defined according to the 1987 American College of Rheumatology criteria), and an age- and sex-matched non-RA cohort. All subjects were followed longitudinally through their complete community medical records, until death, migration, or 1 January 2001. Clinical CV risk factors and outcomes were defined using validated criteria. The chi2 test was used to compare the frequency of each CV risk factor at baseline. Person-years methods were used to estimate the rate of occurrence of each CV risk factor during follow-up. Cox models were used to examine the influence of CV risk factors on the development of CV outcomes. RESULTS: A total of 603 RA and 603 non-RA subjects (73% female; mean age 58 years) were followed for a mean of 15 and 17 years (total: 8842 and 10,101 person-years), respectively. At baseline, RA subjects were significantly more likely to be former or current smokers when compared to non-RA subjects (p<0.001). Male gender, smoking, and personal cardiac history had weaker associations with CV events among RA subjects, compared to non-RA subjects. There was no significant difference between RA and non-RA subjects in the risk imparted with respect to the other CV risk factors (ie, family cardiac history, hypertension, dyslipidaemia, body mass index, or diabetes mellitus). CONCLUSION: While some traditional CV risk factors imparted similar risk among RA compared with non-RA subjects, others (ie, male gender, smoking and personal cardiac history) imparted significantly less risk for the development of CV disease. These differences in the overall impact of traditional CV risk factors suggest that strategies to prevent CV disease and mortality focused solely on controlling traditional CV risk factors may be relatively less beneficial in RA subjects than in the general population. Further research is needed to determine optimal approaches to reducing CV morbidity and mortality in persons with RA.

Raised erythrocyte sedimentation rate signals heart failure in patients with rheumatoid arthritis.

BACKGROUND: Inflammatory markers are associated with heart failure. Patients with rheumatoid arthritis have twice the risk of heart failure compared with people without rheumatoid arthritis. OBJECTIVE: To assess whether heart failure in patients with rheumatoid arthritis is preceded by an inflammatory activation as shown by erythrocyte sedimentation rate (ESR), a systemic marker of inflammation. METHODS: A population-based inception cohort of 575 patients with rheumatoid arthritis, free of heart failure at their rheumatoid arthritis incidence date, was followed up longitudinally until death or 2001. During 15 years of follow-up, they had a median of 15 ESR tests, and 172 patients had new-onset heart failure (Framingham Heart Study criteria). The follow-up period, beginning with the rheumatoid arthritis incidence date and ending with date of the last follow-up, was divided into 6-month intervals. The proportions of patients with at least one ESR value >/=40 mm/h and with anaemia (haemoglobin <11 g/dl) within each 6-month interval were plotted against time from fulfilment of heart failure criteria. A binomial test was used to compare proportions. RESULTS: In patients with rheumatoid arthritis who developed heart failure, the proportion with ESR >/=40 mm/h was highest (23%) during the 6-month period immediately preceding the new-onset heart failure, as compared with the average ESR during the entire remaining follow-up period, both before and after heart failure (10.6%; p<0.01). The proportion of patients with anaemia peaked (54%) during the 6-month period after heart failure. CONCLUSIONS: Inflammatory stimuli may be involved in the initiation of heart failure among patients with rheumatoid arthritis.

Phase II trial of pirfenidone in adults with neurofibromatosis type 1.

We performed an open-label phase II trial of oral pirfenidone in 24 patients with neurofibromatosis type 1 (NF1). Tumors were monitored by three-dimensional MRI. At the end of treatment, four patients had a decrease in tumor volume by 15% or more, three had tumor progression, and 17 remained stable. Pirfenidone warrants further investigation in NF1, which has until now lacked an effective control therapy.

Effect of abciximab on the outcome of emergency coronary artery bypass grafting after failed percutaneous coronary intervention.

OBJECTIVE: To evaluate the outcome of coronary artery bypass grafting (CABG) for failed percutaneous coronary intervention (PCI) in patients who had received abciximab. PATIENTS AND METHODS: In this retrospective study, we analyzed the records of patients who had PCI at our institution between January 1994 and December 1998 and identified those who had urgent or emergency CABG within 48 hours after PCI. CABG was performed for failed PCI in patients who had ongoing ischemia, hemodynamic compromise, or both. These patients were categorized into 2 groups depending on whether they had been given abciximab during PCI. We compared blood product transfusion requirements, bleeding complications, and frequency of in-hospital adverse events of the 2 groups. RESULTS: Of 5636 patients who had PCI, 77 (1.4%) had urgent or emergency CABG within 48 hours, including 11 who were given abciximab (abciximab group) during PCI and 66 who were not given abciximab (no abciximab group). The 2 groups had similar baseline characteristics. The mean +/- SD time to surgery was 8.4 +/-8.0 hours (median, 6 hours) for the abciximab group vs 12.1 +/- 12.5 hours (median, 4 hours) for the no abciximab group. Major bleeding (Thrombolysis in Myocardial Infarction criteria) occurred in 9 (90%) of 10 patients in the abciximab group vs 48 (77%) of 62 patients in the no abciximab group. The total volumes of intraoperative autotransfusion and transfusion of red blood cells and fresh frozen plasma tended to be higher for the abciximab group. Also, this group received a mean of 13.9 U of platelets vs 3.2 U for the no abciximab group (P<.001). However, no in-hospital deaths occurred among patients in the abciximab group, and adverse events were infrequent and comparable between the 2 groups. No difference was noted between the 2 groups in the frequency of surgical reexploration for bleeding. CONCLUSION: Transfusion requirements are higher for patients who undergo emergency or urgent CABG after having received abciximab during PCI. However, in-hospital adverse events are infrequent and comparable to those for patients who do not receive abciximab.

Fas/Fas ligand system mediates epithelial injury, but not pulmonary host defenses, in response to inhaled bacteria.

The Fas/Fas ligand (FasL) system has been implicated in alveolar epithelial cell apoptosis during pulmonary fibrosis and acute respiratory distress syndrome. However, Fas ligation can also lead to cell activation and cytokine production. The goal of this study was to determine the role of the Fas/FasL system in host defenses against Escherichia coli, Staphylococcus aureus, and Streptococcus pneumoniae. We administered bacteria by aerosolization into the lungs of Fas-deficient (lpr) mice and wild-type (C57BL/6) mice and measured bacterial clearance at 6 and 12 h. One hour prior to euthanasia, the mice received an intraperitoneal injection of human serum albumin (HSA) for alveolar permeability determinations. At all times after bacterial challenges, the lungs of the lpr mice contained similar or lower numbers of bacteria than those of the C57BL/6 mice. Alveolar permeability changes, as determined by bronchoalveolar lavage fluid HSA concentrations, were less severe in the lpr mice 6 h after the challenges. In response to E. coli, the lpr mice had significantly more polymorphonuclear leukocytes (PMN) and macrophage inflammatory protein 2 in the lungs, whereas histopathologic changes were less severe. In contrast, in response to the gram-positive cocci, the lpr animals had similar or lower numbers of PMN. We conclude that the Fas/FasL system contributes to the development of permeability changes and tissue injury during-gram negative bacterial pneumonia. The Fas/FasL system did not have a major role in the clearance of aerosolized bacteria from the lungs at the bacterial doses tested.

Recombinant human Fas ligand induces alveolar epithelial cell apoptosis and lung injury in rabbits.

This study investigated whether recombinant human soluble Fas ligand (rh-sFasL) induces apoptosis of primary type II pneumocytes in vitro and lung injury in vivo. Type II cells isolated from normal rabbit lung expressed Fas on their surface and became apoptotic after an 18-h incubation with rh-sFasL. Fas expression in normal rabbit lungs was localized by immunohistochemistry to alveolar and airway epithelia and alveolar macrophages. The administration of 10 microg of rh-sFasL into the right lungs of rabbits resulted 24 h later in both significantly more bronchoalveolar lavage fluid total protein and significantly more tissue changes compared with those in the left lungs, which received rh-sFasL plus Fas:Ig (a fusion protein that binds and blocks sFasL). Tissue changes included thickening of the alveolar walls, neutrophilic infiltrates, apoptotic (terminal deoxynucleotidyltransferase-mediated dUTP nick end labeling-positive) cells in the alveolar walls, and increased expression of interleukin-8 by alveolar macrophages (as determined by immunohistochemistry). We conclude that the alveolar epithelium of normal rabbits expresses Fas and that sFasL induces lung injury and inflammation in rabbits.