Model selection for survival individualized treatment rules using the jackknife estimator.

BACKGROUND: Precision medicine is an emerging field that involves the selection of treatments based on patients' individual prognostic data. It is formalized through the identification of individualized treatment rules (ITRs) that maximize a clinical outcome. When the type of outcome is time-to-event, the correct handling of censoring is crucial for estimating reliable optimal ITRs. METHODS: We propose a jackknife estimator of the value function to allow for right-censored data for a binary treatment. The jackknife estimator or leave-one-out-cross-validation approach can be used to estimate the value function and select optimal ITRs using existing machine learning methods. We address the issue of censoring in survival data by introducing an inverse probability of censoring weighted (IPCW) adjustment in the expression of the jackknife estimator of the value function. In this paper, we estimate the optimal ITR by using random survival forest (RSF) and Cox proportional hazards model (COX). We use a Z-test to compare the optimal ITRs learned by RSF and COX with the zero-order model (or one-size-fits-all). Through simulation studies, we investigate the asymptotic properties and the performance of our proposed estimator under different censoring rates. We illustrate our proposed method on a phase III clinical trial of non-small cell lung cancer data. RESULTS: Our simulations show that COX outperforms RSF for small sample sizes. As sample sizes increase, the performance of RSF improves, in particular when the expected log failure time is not linear in the covariates. The estimator is fairly normally distributed across different combinations of simulation scenarios and censoring rates. When applied to a non-small-cell lung cancer data set, our method determines the zero-order model (ZOM) as the best performing model. This finding highlights the possibility that tailoring may not be needed for this cancer data set. CONCLUSION: The jackknife approach for estimating the value function in the presence of right-censored data shows satisfactory performance when there is small to moderate censoring. Winsorizing the upper and lower percentiles of the estimated survival weights for computing the IPCWs stabilizes the estimator.

How do pig farms maintain low Salmonella prevalence: a case-control study.

Salmonella prevalence in UK pigs is amongst the highest in Europe, highlighting the need to investigate pig farms which have managed to maintain a low Salmonella seroprevalence. A total of 19 pig farms that had a consistently low (<10%) seroprevalence over 4 years (named Platinum farms) were compared against 38 randomly selected Control farms, chosen to match the same distribution of production types and geographical distribution of the Platinum farms. Each farm was visited and floor faeces and environmental samples were collected. It was shown that Control farms had a significantly higher median percentage of pooled faecal samples positive for Salmonella compared with the Platinum farms (12.1% and 0.4% for pooled faecal samples, respectively) and were more likely to have serovars of public health importance detected (S. Typhimurium/ monophasic variants or S. Enteritidis). Considering the comprehensive on-farm sampling, the identification of farms negative for Salmonella, along with the identification of those that had maintained low prevalence over a long period is important. The risk factor analyses identified pelleted feed, feed deliveries crossing farm perimeter and regular antibiotic use as associated with being a Control farm. Performance data indicated that Platinum farms were performing better for slaughter live weight than Controls. Limited assessments of available pig movement records suggested that the source of pigs was not key to Platinum status, but further study would be needed to confirm this finding. These results emphasise that maintaining very low prevalence on UK farms is achievable.

Phase II study of gemcitabine and bexarotene (GEMBEX) in the treatment of cutaneous T-cell lymphoma.

BACKGROUND: Both gemcitabine and bexarotene are established single agents for the treatment of cutaneous T-cell lymphoma (CTCL). We investigated the feasibility and efficacy of combining these drugs in a single-arm phase II study. METHODS: Cutaneous T-cell lymphoma patients who had failed standard skin-directed therapy and at least one prior systemic therapy were given four cycles of gemcitabine and concurrent bexarotene for 12 weeks. Responders were continued on bexarotene maintenance until disease progression or unacceptable toxicity. RESULTS: The median age was 65 years, stage IB (n=5), stage IIA (n=2), stage IIB (n=8), stage III (n=8) and stage IVA (n=12), 17 patients were erythrodermic, 17 patients were B1, and 10 patients were both erythrodermic and B1. Thirty (86%) patients completed four cycles of gemcitabine. In all, 80.0% of patients demonstrated a reduction in modified Severity-Weighted Assessment Tool (mSWAT) score although the objective disease response rate at 12 weeks was 31% (partial response (PR) 31%) and at 24 weeks 14% (PR 14%, stable disease (SD) 23%, progressive disease (PD) 54%, not evaluable 9%). Median progression-free survival was 5.3 months and median overall survival was 21.2 months. CONCLUSION: The overall response rate of the combination did not reach the specified target to proceed further and is lower than that previously reported for gemcitabine as a single agent.

Breast cancer risk following Hodgkin lymphoma radiotherapy in relation to menstrual and reproductive factors.

BACKGROUND: Women treated with supradiaphragmatic radiotherapy (sRT) for Hodgkin lymphoma (HL) at young ages have a substantially increased breast cancer risk. Little is known about how menarcheal and reproductive factors modify this risk. METHODS: We examined the effects of menarcheal age, pregnancy, and menopausal age on breast cancer risk following sRT in case-control data from questionnaires completed by 2497 women from a cohort of 5002 treated with sRT for HL at ages <36 during 1956-2003. RESULTS: Two-hundred and sixty women had been diagnosed with breast cancer. Breast cancer risk was significantly increased in patients treated within 6 months of menarche (odds ratio (OR) 5.52, 95% confidence interval (CI) (1.97-15.46)), and increased significantly with proximity of sRT to menarche (Ptrend<0.001). It was greatest when sRT was close to a late menarche, but based on small numbers and needing reexamination elsewhere. Risk was not significantly affected by full-term pregnancies before or after treatment. Risk was significantly reduced by early menopause (OR 0.55, 95% CI (0.35-0.85)), and increased with number of premenopausal years after treatment (Ptrend=0.003). CONCLUSION: In summary, this paper shows for the first time that sRT close to menarche substantially increases breast cancer risk. Careful consideration should be given to follow-up of these women, and to measures that might reduce their future breast cancer risk.

UK-wide risk factor study of broiler carcases highly contaminated with Campylobacter.

Campylobacter is a major cause of food poisoning and is typically the most common cause of gastroenteritis in the UK. Data collected at broiler farm and abattoir level, for slaughter batches that were sampled for UK-wide monitoring, were used to generate two epidemiological risk factor models. A total of 483 batches slaughtered between January 2016 and March 2017 were used in the analysis, coming from 19 abattoirs representing more than 85% of UK broiler production. For each selected slaughter batch, one carcase was sampled after primary chilling and 10 randomly sampled birds had caecal samples collected at the evisceration point. Samples were used for Campylobacter identification and quantification. Two multivariable mixed-effects models were designed, one with the binary outcome for the detection of a highly contaminated (>1000 colony forming units (CFU)/g) Campylobacter-positive carcase, whereas the other used the Campylobacter colony count (CFU/g) carcase outcome. The results suggest that caecal colonization within the batch was a key factor for the occurrence of Campylobacter on carcases, and many factors that were identified in the model were also likely to be related to colonization or related to the risk of introduction of Campylobacter from partial depopulation (referred to as thinning) of ~30% of the flock approximately 1 week before full flock depopulation events. The amount of neck skin in the sample was another key factor identified and was included in both models as a risk factor. The models have also identified other factors which may be related to the general health and husbandry on-farm (use of prebiotics or vaccines, and identification of the product used for drinking line cleaning), whereas the other factors may identify control points related to transmission within a farm. The identification of these variables could help focus control efforts on-farm, especially for relatively easy improvements, such as improving the provision of house-specific bird-weighing buckets/cages in houses.

Cultural Transformation in Healthcare: How Well Does the Veterans Health Administration Vision for Whole Person Care Fit the Needs of Patients at an Academic Rehabilitation Center?

Background: The Veterans Health Administration is undergoing a cultural transformation toward person-driven care referred to as the Whole Health System of Care. Objective: This pilot study evaluated whether the Whole Health model resonates with patients of a large public university rehabilitation clinic. Methods: Thirty participants completed the Veterans Health Administration's Personal Health Inventory (PHI), and six attended the course "Taking Charge of My Life and Health." Researchers analyzed PHI responses and post-course focus group transcripts. A short post-PHI survey and post-course evaluation were collected. Results: Participants agreed the PHI is a simple, useful tool. The course, while well attended, did not meet participants' expectations. Participants wanted access to integrative therapies and opportunities to contribute to healthcare transformation. Conclusion: Rehabilitation patients resonated with the Whole Health vision. They expressed enthusiasm for the cultural transformation represented by the model along with frustration that standard healthcare experiences fall short of this vision.

Determining pig holding type from British movement data using analytical and machine learning approaches.

Determining the size, location and structure of a livestock population is an essential aspect of surveillance and research as it provides understanding of the representativeness and coverage of any project or scheme. It is an important input for a variety of epidemiological analyses, for example, allowing generation of more accurate sample size calculations for estimating prevalence or freedom from disease, cost-benefit analyses for control measures to reduce or eradicate livestock disease, or development of between-herd network models to estimate the impact of movement of animals between farms on the spread of livestock diseases. The work described here provides information on how British pig movement data was compared against other datasets related to the British pig population to define its appropriateness for defining pig holding demographics. The data were then used to identify the location of pig holdings and the estimated herd size (split into five categories). Two methods are described that were used to classify the holding type of the identified pig holdings. The first method was an epidemiological method that used expert opinion to determine a set of rules based on movement characteristics to classify each holding. The second method was a machine learning approach that used k means cluster analysis to automatically estimate the holding type based on a set of proxy indicators. Each method had a good accuracy rate, when compared to matched holdings present in data provided by the Annual June Agricultural Survey, but all misclassified some holdings. While both of the methods on their own provided a reasonable estimate, it was concluded that a consensus model, considering the results of both models and the Survey, provided the most accurate result. However, the machine learning approach was beneficial, as although some technical expertise was needed to set up the model, it was considerably faster to implement than the other method, as well as being quicker and easier to adapt and re-run with updated information.

Quality Assurance Peer Review for Radiotherapy for Haematological Malignancies.

AIMS: To assess the impact of weekly scheduled peer review of radiotherapy planning contours for definitive treatment of haematological malignancies based on rates of recommended changes. MATERIALS AND METHODS: Analysis of a prospective database of contour-based peer review at weekly scheduled meetings for patients undergoing definitive radiotherapy for haematological malignancies at a single large cancer centre between January and December 2018. Recommended changes were prospectively classified as involving the gross tumour volume (GTV), clinical target volume (CTV), planning target volume (PTV), organs at risk or dose fractionation. A univariate analysis was carried out to explore the associations between recommended changes and disease, treatment characteristics and consultant experience. RESULTS: In total, 158/171 (92%) of all cases of haematological malignancy undergoing definitive radiotherapy were prospectively peer reviewed over a 12-month period. Overall, 26/158 (16.5%) changes were recommended within the peer review meetings. This included a total of 27 contour changes (GTV, CTV or PTV) in 25 patients. An increase in CTV was the most common change, occurring in 20/158 (12.7%) cases. One dose-fractionation change was recommended. Additional advice regarding planning technique/set-up was documented in 5/158 (3.2%) patients. There were no significant associations between rates of recommended change and disease type, stage, prior chemotherapy, first line versus refractory/relapse, anatomical site, radiotherapy technique or consultant experience. CONCLUSIONS: Weekly contour-based peer review meetings resulted in a high rate of recommended changes. Compliance was high. Peer review was potentially beneficial for all disease and treatment characteristics and for any degree of clinician experience.

Intra-abdominal fat measurement by ultrasonography: association with anthropometry and metabolic syndrome in adolescents.

OBJECTIVE: To associate intra-abdominal fat thickness measured by ultrasonography to the factors related to metabolic syndrome and to determine cutoff points of intra-abdominal fat measurement associated with a greater chance of metabolic syndrome in adolescents. METHODS: This was a cross-sectional study, with 423 adolescents from public schools. Intra-abdominal fat was measured by ultrasonography. Anthropometric data were collected, and biochemical analyses were performed. RESULTS: Intra-abdominal fat was measured by ultrasonography, showing a statistically significant association with the diagnosis of metabolic syndrome (p=0.037), body mass index (p<0.001), elevated triglyceride levels (p=0.012), decreased plasma HDL levels (p=0.034), and increased systemic blood pressure values (p=0.023). Cutoff values of intra-abdominal fat thickness measurements were calculated by ultrasound to estimate the individuals most likely to develop metabolic syndrome. In the logistic regression models, the cutoff values that showed the highest association with metabolic syndrome in males were 4.50, 5.35, 5.46, 6.24, and 6.50cm for the ages of 14, 15, 16, 17, and 18/19 years, respectively. In the female gender, the cutoff values defined for the same age groups were 4.46, 4.55, 4.45, 4.90, and 6.46cm. In an overall analysis using the ROC curve, without gender and age stratification, the cut-off of 3.67cm showed good sensitivity, but low specificity. CONCLUSION: Ultrasonography is a useful method to estimate intra-abdominal adipose tissue in adolescents, which is associated with the main factors related to obesity and metabolic syndrome.

Long-term results of a randomised trial of involved field radiotherapy vs extended field radiotherapy in stage I and II Hodgkin lymphoma.

Involved field (IF) radiation was compared with extended field (EF) radiation in Hodgkin lymphoma (HL) to ascertain whether reduced radiation fields would reduce the late sequelae of radiation without compromising disease control and survival. A total of 603 patients with stage I or II HL were entered into this trial; laparotomy was carried out in 380 (63%) patients. Stage I or IIA disease patients were randomised to receive IF or EF comprising a mantle or inverted Y fields alone. Stage I and IIB patients were randomised between mantle or inverted Y fields and total nodal irradiation (TNI). The dose was 35 Gy to uninvolved sites and 40 Gy to involved sites. The median followup of surviving patients was 25.2 years with only 3.3% lost to follow-up. The treatment failure rate at 25 years in stage IA and IIA was 44% after EF and 54% after IF (P = 0.01); in stage I and IIB this was 80% (EF) and 82% (TNI) at 25 years. No difference in overall survival between the randomised groups was seen. The incidence of second malignancies was 21% after IF and 20% after EF with a slight excess of lung cancer in the EF group. No significant differences in the causes of death between the randomised arms have emerged. In conclusion, IF radiotherapy for stage I and IIA HL results in a 11% greater risk of relapse compared with EF but has no effect on overall survival, risk of second malignancy or cause of death at 25 years.

The Minimum Competences in Non-surgical Oncology that Medical Students Need to Acquire in Order to be Safe Foundation Year 1 (F1) Doctors: A Delphi Survey.

AIMS: To develop a consensus on the minimum competences in non-surgical oncology that medical students need to acquire in order to be safe Foundation Year 1 (F1) doctors. MATERIALS AND METHODS: A two-round Delphi survey was conducted by e-mail with an expert panel of 24 consultant oncologists who had expressed an interest in undergraduate education. RESULTS: The response rate to round 1, which asked panellists to list the competences they thought were important, was 50%. The competences they generated contained 86 different concepts. These were categorised according to the learning outcomes in Tomorrow's Doctors. The panellists were then asked to rate the importance of each proposed competence between 1 and 9 on a Likert scale to give a measure of the perceived importance and consensus. The panellists generated competences in all the main categories of learning outcomes in Tomorrow's Doctors. The scores were highest and the consensus greatest for those competences related to the doctor as a practitioner and the doctor as a professional. CONCLUSION: The Delphi survey was an effective method of obtaining the judgement of an expert panel and in measuring the degree of consensus. The results of the survey were valuable in informing the design of a UK non-surgical oncology curriculum.

Involved Node, Site, Field and Residual Volume Radiotherapy for Lymphoma: A Comparison of Organ at Risk Dosimetry and Second Malignancy Risks.

AIMS: Recent radiotherapy guidelines for lymphoma have included involved site radiotherapy (ISRT), involved node radiotherapy (INRT) and irradiation of residual volume after full-course chemotherapy. In the absence of late toxicity data, we aim to compare organ at risk (OAR) dose-metrics and calculated second malignancy risks. MATERIALS AND METHODS: Fifteen consecutive patients who had received mediastinal radiotherapy were included. Four radiotherapy plans were generated for each patient using a parallel pair photon technique: (i) involved field radiotherapy (IFRT), (ii) ISRT, (iii) INRT, (iv) residual post-chemotherapy volume. The radiotherapy dose was 30 Gy in 15 fractions. The OARs evaluated were: breasts, lungs, thyroid, heart, oesophagus. Relative and absolute second malignancy rates were estimated using the concept of organ equivalent dose. Significance was defined as P < 0.005. RESULTS: Compared with ISRT, IFRT significantly increased doses to lung, thyroid, heart and oesophagus, whereas INRT and residual volume techniques significantly reduced doses to all OARs. The relative risks of second cancers were significantly higher with IFRT compared with ISRT for lung, breast and thyroid; INRT and residual volume resulted in significantly lower relative risks compared with ISRT for lung, breast and thyroid. The median excess absolute risks of second cancers were consistently lowest for the residual technique and highest for IFRT in terms of thyroid, lung and breast cancers. The risk of oesophageal cancer was similar for all four techniques. Overall, the absolute risk of second cancers was very similar for ISRT and INRT. CONCLUSIONS: Decreasing treatment volumes from IFRT to ISRT, INRT or residual volume reduces radiation exposure to OARs. Second malignancy modelling suggests that this reduction in treatment volumes will lead to a reduction in absolute excess second malignancy. Little difference was observed in second malignancy risks between ISRT and INRT, supporting the use of ISRT in the absence of a pre-chemotherapy positron emission tomography scan in the radiotherapy treatment position.

The response of a rodent fibrosarcoma to superficial/interstitial photochemotherapy, chemotherapy or radiotherapy.

Growth and dose-response curves were established for a subcutaneously implanted isogenic fibrosarcoma in BD9 rats after treatment with photochemotherapy (PCT), using Photofrin II or polyhaematoporphyrin with superficial or interstitial 630 nm light, cyclophosphamide or gamma-irradiation. Tumour response to PCT increased with dose up to 200 J.cm-2 for superficial light or 200 J for interstitial light but no further response occurred after higher light doses. The maximum response after interstitial treatment was significantly greater than after superficial treatment where only a small margin of normal tissue was treated. The incidence of necrosis in the overlying skin was significantly less after interstitial than superficial light suggesting a better therapeutic ratio after interstitial than superficial PCT. Tumour response increased with the diameter of the treatment field after superficial light supporting the possibility of a tumour bed effect associated with PCT. The largest tumour that could be effectively treated with a single optical fibre was 12 mm. The dose-response curves for interstitial PCT and cyclophosphamide were similar but ionizing irradiation produced increasing tumour response throughout the range of doses used (5 to 30 Gy) and the maximum response was greater after radiotherapy than after PCT or chemotherapy suggesting that in this tumour model interstitial PCT is as effective as cyclophosphamide but less effective than radiotherapy.

Identification of the major T-cell antigens present in the Brucella melitensis B115 protein preparation, Brucellergene OCB.

Brucellergene is a commercial allergen prepared from Brucella melitensis strain B115 and containing at least 20 cytoplasmic proteins. These proteins were separated by SDS-PAGE. The unstained gel was divided into 18 fractions and proteins were eluted from the gel fractions. The capacity of the separated proteins to elicit delayed-type hypersensitivity (DTH) in infected guinea-pigs or to induce the production of interferon-gamma (IFN-gamma) by blood cells from infected cattle was evaluated. The biological activity of the corresponding protein fractions blotted on to nitrocellulose was measured in a lymphocyte blastogenesis assay. Among the 18 fractions tested, two-spanning the mol. wt ranges 17-22 (fraction 8) and 35-42-kDa (fraction 17)-showed the maximum biological activity in the three tests. These fractions contain two antigens, the Brucella bacterioferritin (BFR) and P39 proteins. Both proteins are good candidates for the detection of cellular immunity to Brucella.

The development of targeted chemotherapy for CNS lymphoma-a pilot study of the IDARAM regimen.

PURPOSE: We have developed and evaluated a CNS-targeted chemotherapy regimen based on the pharmacokinetic properties of the individual drugs in the combination. PATIENTS AND METHODS: In a twin-track study, 16 patients with secondary CNS lymphoma (SCNSL) and 8 with primary CNS lymphoma (PCNSL) were treated with IDARAM which comprised idarubicin 10 mg/m(2) i.v., days 1 and 2; dexamethasone 100 mg, 12-h infusion, days 1, 2 and 3; cytosine arabinoside (ARA-C) 1.0 g/m(2), 1-h infusion, days 1 and 2; methotrexate 2.0 g/m(2), 6-h infusion, day 3 (with folinic acid rescue); and cytosine arabinoside 70 mg plus methotrexate 12 mg, intrathecally, days 1 and 8. Two cycles were delivered at 3-weekly intervals. After response assessment, patients received adjuvant cranial radiotherapy (40 Gy over 20 fractions). RESULTS: The series comprised 24 patients, 11 male and 13 female. Their median age was 53 years (range 21 to 73 years). Grade 4 neutropenia and thrombocytopenia occurred in the majority of patients treated. Of the eight PCNSL patients, seven achieved complete remission (CR). Four remained in CR at the time of this report with a median duration of follow-up of 25 months (range 11 to 42 months). Of the 16 SCNSL patients, 12 achieved CR. Seven patients remained in CR at the time of this report with a median duration of follow-up of 24 months (range 18 to 57 months). CONCLUSION: This study suggests that IDARAM is an effective regimen in both PCNSL and SCNSL and is suitable for further development and evaluation.

Long term outcome in adolescents with Hodgkin's lymphoma: poor results using regimens designed for adults.

It is unclear whether the outcome in adolescents with Hodgkin's lymphoma is as good as that in children and there are no prospective randomized trials comparing regimes used in children and adults in this setting. We have therefore performed an analysis of 210 adolescent patients diagnosed with Hodgkin's lymphoma between 1970-1997 and registered on the database held by the British National Lymphoma Investigation. Patients were treated according to adult regimens current at the time of their diagnosis. The complete response rate recorded in 209 patients was 76%. This was highly dependent on disease stage being 95% in patients with localized disease but 63% in those with advanced disease. The 5 year event free survival for the whole cohort was 50% falling to 41% at 20 years with overall survival of 81% falling to 68% at 5 and 20 years respectively. There is no significant difference in the 3 decades pertaining to this analysis. Of the 62 deaths in this cohort, 70% were due to Hodgkin's lymphoma but of the 13 deaths occurring beyond 10 years, only 3 were due to Hodgkin's lymphoma, the reminder being attributable to the late effects of therapy. Results from paediatric groups have been much more encouraging than those presented from this cohort. It seems the use of risk-adjusted combined modality therapy with minimization of radiation fields and doses and reduction of anthracycline and alkylator exposure has been successful in children and should be used in adolescents.