Using artificial intelligence to bring evidence-based medicine a step closer to making the individual difference.

The vision of evidence-based medicine is that of experienced clinicians systematically using the best research evidence to meet the individual patient's needs. This vision remains distant from clinical reality, as no complete methodology exists to apply objective, population-based research evidence to the needs of an individual real-world patient. We describe an approach, based on techniques from machine learning, to bridge this gap between evidence and individual patients in oncology. We examine existing proposals for tackling this gap and the relative benefits and challenges of our proposed, k-nearest-neighbour-based, approach.

A prospective comparison of multidisciplinary treatment of oesophageal cancer with curative intent in a UK cancer network.

AIMS: Combined modality therapy (with chemotherapy+/-radiotherapy) has become a standard treatment for locally advanced oesophageal cancer. However, there appears to be no compelling evidence for one treatment type or combination to suit all and at this time the clinical multi-disciplinary team (MDT) forms an important role in selecting optimal therapies for the individual. This prospective comparison in one cancer network, looks at the outcomes of this decision making process. METHODS: Over a five year period 1998-2003, data were prospectively collected on all 330 consecutive patients, referred to a tertiary specialised MDT for whom curative treatment was the planned intent. Patients were managed according to an agreed local protocol and allocated to receive one of 5 treatments: surgery alone (S), pre-operative chemotherapy (C+S), pre-operative chemo-radiotherapy (CRT+S), definitive chemo-radiotherapy (CRT) and radiotherapy alone (RT). RESULTS: The 2 and 5 year survival for all patients receiving potentially curative treatment were 49% and 26% respectively. With 2 and 5 year survival for S, CRT+S, C+S, CRT and RT being 53,21; 57,40; 37,27; 50,27; 23,0 months respectively. Of the surgical therapies, mortality was highest in the CRT+S group, versus C+S and S; 12.5%, 1.6%, 4.5% respectively (p=0.025). Non-surgical based therapies had more than double the incidence of local relapses compared to surgical based therapies; however the CRT group had an overall survival comparable with S alone. The commonest sites of distant relapse were liver (56%), lung (38%), bone (32%) and non-regional lymph nodes (24%). CONCLUSION: The results suggest that in patients who are deemed unfit for surgical intervention, definitive chemoradiotherapy remains a viable alternative; they also lend further support to selected case triple modality therapy. These areas should be further examined in the context of randomised controlled phase III trials.

Small cell lung cancer treated in southeast Wales.

AIMS: In small cell lung cancer (SCLC), consolidation thoracic irradiation (CTI) increases 3-year absolute survival by 5.4% in patients with limited disease and a complete response to chemotherapy. Early concurrent thoracic radiotherapy has been shown to improve local control and prolong survival compared with CTI in some trials. The standard management of patients with SCLC in southeast Wales is CTI in individuals with limited disease and a complete response to chemotherapy. A review of patients with SCLC was carried out to establish whether survival locally is comparable with that reported in published studies, and if patients given CTI have survival comparable with that reported in studies where early concurrent thoracic radiotherapy was used. MATERIALS AND METHODS: Between January 2000 and December 2002, 303 patients were registered with SCLC in southeast Wales. One hundred and fifteen (47%) patients had limited disease and 60/115 (52%) received CTI. RESULTS: Patients with limited disease receiving CTI had a median survival of 17.7 months (95% confidence interval: 15-27.9 months). The 2- and 5-year survivals were 38 and 13%, respectively. CONCLUSIONS: These results compare favourably with previously published studies on SCLC. There are no plans to change our current treatment policy for SCLC in southeast Wales.

Hypofractionated radiotherapy for muscle invasive bladder cancer in the elderly.

BACKGROUND AND PURPOSE: We have retrospectively investigated a hypofractionated regimen in a cohort of 65 elderly patients (median age 78 years), designed to minimise acute radiation affects and maximise patient tolerance and convenience in this frail group. MATERIALS AND METHODS: All patients were CT planned to a small volume. Once weekly fractions (6 Gy) prescribed to the 100% isodose as a target minimum to 30 Gy (n = 53) and 36 Gy (n = 12) were administered. Palliation of symptoms before, during, and 1 month from completion of radiotherapy were graded using the urinary and bowel symptom and toxicity index. RESULTS: Fifty-five patients had persisting urinary symptoms following trans urethral resection of bladder. Twenty-eight (51%) were completely palliated of symptoms and 7 (13%) noticed an improvement at a 1 month review. Ninety-two percent of patients with haematuria were completely palliated compared to only 24% of those with dysuria and frequency. The median symptom free interval was 7 months (range 2-40months). Median overall survival was 9 months (range 2-41months). Twelve percent of patients required inpatient admission and only three failed to complete the prescribed course due to bowel toxicity. Grade 3 acute urinary and bowel treatment related toxicity, were recorded in 18% and 9% of patients, respectively. In total, 43% of patients noticed a transient worsening of their presenting symptoms on treatment. To date no significant late toxicity (>grade 2) has been recorded. CONCLUSIONS: This regimen is generally well tolerated and offers reasonable palliation of symptoms on an outpatient basis for this frail poor prognosis group. Haematuria is particularly well palliated although only a quarter of patients presenting with dysuria and frequency were rendered symptom free.

High-dose folinic acid and 5-fluorouracil bolus and continuous infusion in advanced colorectal cancer: poor response rate in unselected patients.

We have conducted a retrospective study of high-dose folinic acid and 5-fluorouracil in 96 patients with advanced colorectal cancer. Patients received 200 mg m-2 (maximum 300-350 mg) folinic acid by infusion over 2 h followed by an i.v. bolus of 5-fluorouracil 400 mg m-2 then an infusion of 5-fluorouracil 600 mg m-2 over 22 h. This was repeated over the next 24 h. The schedule was given every 2 weeks for four cycles; thereafter patients with objective response continued to a maximum of eight cycles. The overall response rate was 10.6% in 85 evaluable patients. The median duration of response was 11 months. The median survival was 6 months. Toxicity was low, only one patient experiencing toxicity greater than WHO grade II (grade IV platelet toxicity). Diarrhoea, nausea, vomiting and mucositis also occurred but were mild and infrequent. Our low response rate may be related to factors such as patient characteristics or duration of treatment.

The influence of grid size on accuracy in radiotherapy dose plotting.

The recent article by Niemierko and Goitein [Med. Phys. 16, 239-247 (1989)] illustrates well the errors which are occurring in plotting isodose lines. We wish to augment their analysis with similar work done at Cardiff a few years ago; to indicate some practical treatment outcomes they have omitted; to propose even more stringent requirements of the grid size used; and thus to further alert users and software manufacturers to this problem.

Floristic variation in plant communities on metalliferous mining residues in the northern and southern Pennines, England.

Vegetation and soil surveys were conducted on metalliferous mine wastes in the northern and southern Pennines of England. Analyses of vegetation composition in relation to soil chemical variation were performed. Ordination analysis facilitated the detection of groups of co-occurring species which are characteristic of types of metal-contaminated soil. The results of regression analysis implied the importance of soil pH and concentration of available lead or zinc, depending upon region, in determining species distributions. A strong interactive effect was evident between soil phosphorus and zinc content in influencing species distributions in the southern Pennines sample.Regional similarities in the vegetation of apparently similar metalliferous soils existed. The unusually high species richness of some soils was associated with relatively low concentrations of heavy metals. Many species of relatively floristically-rich wastes were also colonists of surrounding grasslands or woodland. This indicated the potential importance of propagule availability and capability for rapid establishment on bare or unstable ground. Mechanisms of physiological stress avoidance, rather than heavy-metai tolerance, may explain the occurrence of non-metallophytes on soils contaminated with lead and zinc.

PRESC: an interactive program for radiation therapy prescriptions.

An interactive program for prescribing radiotherapy treatments has been developed with the aim of improving radiation dose accuracy by preventing mistakes in quantitative data handling and calculations. The program also aims to decrease the time required to prescribe treatments and check the calculations. The interactive dialogue closely follows present, manual methods of prescription. The program provides a comprehensive choice of treatment options which either individually or in combination cover all the treatments carried out on the Cobalt units for which the program was designed. Each of the treatment options has been highly optimized to reduce the amount of data entered and for ease of use. The program allows changes to the prescription at time of entry and during the treatment course. The treatment sheet and patient record resulting from the interactive dialogue are discussed and compared with the existing system. The modular program structure is described and some of the results from use of the program are discussed. Possible future improvements are outlined.

A microcomputer system for prescription, calculation, verification and recording of radiotherapy treatments.

The design of a microcomputer system for the reduction of mistakes in radiotherapy is described. The system covers prescription entry, prescription and treatment calculations, and verification and recording of the treatment set-up. A telecobalt unit was interfaced to the system and in the first 12 months 400 patients have been prescribed and 5000 treatment fields verified. The prescription is entered by the medical officer using an interactive program and this prescription provides the reference for verifying the treatment set-up. The program allows amendments to the prescription to be made easily during the treatment course. The treatment parameters verified are field size, wedge and treatment time. The system uses bar-codes for patient and field identification. A reduction in the number of mistakes has been achieved and future developments are discussed.