Search

Background: A metaanalysis is often performed if risk assessment with available epidemiological data is not sensible. The results of these analyses serve mainly to quantify the risk of weak associations., Material and Methods: In this paper we describe methodologic issues of this approach with emphasis on the difference between metaanalysis from published data and from individual patient records. As an example we discuss studies on oral contraceptives and breast cancer. Limits of metaanalysis from published data are given., Conclusion: Although metaanalysis with individual data may be cost- and time-consuming their results are more reliable.

Metaanalyses of epidemiological studies have increased during the last years and are often used to evaluate the effect of risk factors which are inconsistent in different studies, mainly for small risk factors. Very often a metaanalysis is performed from published data. In this article we discuss this form of a metaanalysis and investigate whether the requirement to get reliable information is achievable with it. We mainly ask questions whether qualitative and quantitative dose-response analysis can be performed. We point out the differences between metaanalysis from experimental data and clinical randomized studies and epidemiological studies. We discuss different arguments that were given for performing metaanalysis in clinical trials and investigate whether they are also valid in observational studies. We mainly concentrate on the problem of estimating a single pooled risk estimate. Two examples from literature are used to show problems with metaanalysis from published data.

Transvaginal sonomorphologie and colour Doppler measurements were obtained preoperatively in 212 adnexal tumours: 81 premenopausal tumours (13 malignant and 68 benign) and 131 postmenopausal tumours (55 vs 76). Tumours were divided into five different scores according to their sonomorphology [16]. Scores I and II are related to benign tumours. Score V represents typically malignant tumours. Scores III and IV are associated with benign and malignant tumours. If score I and II are considered as benign and score III to V as malignant the sensitivity in pre- and postmenopausal tumours is 90%. However, the specificity is only 56% vs 70% respectively. In order to improve the accuracy, colour Doppler was additionally performed in tumours with sonomorphological score III and IV. The following criteria were tested: minimum resistance index (Rlmin), number of tumour arteries (ART), maximum (Smax) and sum (Ssum) of peak systolic velocities. All criteria showed significant differences between benign and malignant tumours. Tumours of score III and IV were differentiated by colour Doppler with an accuracy between 66% and 81% for premenopausal and 69% to 86% for postmenopausal women. The combination of sonomorphology and colour Doppler increased the accuracy between 84% and 90% with a sensitivity of up to 92% in pre- and 89% in postmenopausal patients. Sequential colour Doppler sonography as a supplement to transvaginal sonography improves tumour differentiation. The limitation of colour Doppler measurements to score III and IV lesions reduced the length of examination time to a reasonable extent.

The University of Erlangen has been engaged in clinical obstetrics for approximately 170 years. During this time, Erlangen University's delivery house, opened in 1828 and at first having considerably less than 50 births a year, developed into a perinatal centre with approximately 1,700 births a year. For the period from 1880 to 1981, a group of MD students reviewed the existing records and evaluated 60,000 births with respect to more than 40 parameters. Part of the results obtained are shown with special reference to operative obstetrics. Apart from the general influence of the scientific development on decisions and results within obstetrics, individual factors were also recognizable, factors which are linked with the experiences, insights and specialized working areas of the particular head of the hospital.

In 129 patients with ovarian tumours, a preoperative examination by transvaginal colour Doppler sonography (ATL UM9 HDI) was performed. 45 were malignant and 84 benign. 46 patients were premenopausal and 83 postmenopausal. All tumour vessels were analysed by spectrum analysis. The number of tumour vessels and the lowest resistance index (RImin) and pulsatility index (PImin) of all vessels were evaluated as diagnostic criteria of the neovascularisation. Up to 19 vessels were found in 115 (89%) tumours. In benign lesions, the number of vessels was significantly lower than in malignancies (median 3 versus 9, p < 0.0001). The RImin and PImin values between benign and malignant tumours were also significantly different, independent of the menopausal status. In postmenopausal patients, the median RImin in 42 benign tumours was 0.62 (0.26-1.0) and in 41 malignancies 0.40 (0.22-0.66). The median PImin was 0.98 (0.31-5.84) in benign tumours and 0.53 (0.25-1.22) in malignant tumours. However, due to the overlapping range between benign and malignant tumours, these parameters do not allow an accurate differentiation. With a cut off value for RImin = 0.5 and PImin = 0.7, the sensitivity is 85% or 82.5% respectively. The specificity for RImin and PImin is 77%, and the diagnostic accuracy is 84% and 80% respectively. The correlation coefficient between RImin and PImin is 0.99 in benign and malignant tumours and both indices are adequate. The calculation of the RI is more easily available and should be preferred as a diagnostic parameter.

Between 1979 and 1986, 74 patients with hypopharyngeal carcinomas were operated using transoral laser microsurgery by the first author. 32 of the patients were subdivided into 5 subgroups and considered separately because of pretreatment for head and neck tumors, simultaneous multiple tumors etc. (excluding criterias). Survival times were not significantly prolonged and lasted 1-27 months (median, 11 months), but the quality of life was improved due to preservation or restoration of natural laryngopharyngeal functions. Forty-two patients were operated with curative intention. This group primarily underwent transoral laser microsurgery, aiming at complete locoregional tumor resection with function preservation (pT1, 5; pT2, 31; pT3, 4; pT4, 2). In 29 patients 31 necks were operated, mostly as a regionally limited functional neck dissection (or "selective" neck dissection). In 90% of the cases neck metastases (pN+) were found, mostly in levels II and III; pN1, 6; pN2a, 1; pN2b, 18; pN2c, 1. Altogether, stages III and IV were found in 71.4% of the patients. A temporary tracheotomy was required in four patients. There was no secondary laryngectomy, even though it was indicated in one case. Post-treatment oncological followup (median observation time, 104 months) demonstrated loco-regional recurrences (n = 1), late or recurrent metastases (n = 4), persisting metastases in the neck with cerebral metastasis (n = 1), distant metastases (n = 4), secondary tumors (n = 9, 5 of which occurred in the head and neck). Through March 1993, 24 patients (57%) have died. Causes were TNM-related (7), second primary tumor with or without distant metastases (8) and intercurrent disease with no evidence of disease (9). Within 5 years 17% of the patients died of TNM-related tumors, 9.5% due to a second primary with or without distant metastases, as well as 9.5% with intercurrent disease. The 5-year overall survival rate was 64% and was 83% (adjusted survival rate) if only TNM-related deaths were considered.

In the German Breast Cancer Study Group (GBSG) multicenter trial "Breast preservation in small breast cancer" 1119 patients were accrued over a period of six years. The study was originally designed as a randomized trial which had to be changed into a prospective observation study. A high therapeutic standard was guaranteed by quality control. The univariate analysis of prognostic variables was the first step to a valid treatment comparison. Those factors determined as being significant were combined with the treatment effects in a multivariate analysis. The published results were obtained after a median observation time of four years.

Besides characteristic morphological differences, carcinomas of the breast show remarkable changes of vascularity, which are essential for their enhanced metabolism. High-frequency CW Doppler allows to investigate normal and pathological vascularisation in the breast. In our basic study we examined 200 patients, 37 with breast malignancies (6 non-palpable lesions). 33 of the carcinomas showed pathological vascularisation. In an extended examination procedure we investigated also normal breast vascularisation and compared it with blood flow in lesions. We developed therefrom an easy and rational method which allows physiological blood flow measurements as well as a differentiation of breast lesions within a few minutes' time.

There is no method that allows satisfactory differentiation between benign and malignant breast lesions. We were able to differentiate 94% of palpable malignant tumours with CW-doppler ultrasound, which helps to increase the efficiency of clinical examinations, breast sonography and mammography. Two hundred patients were examined with an 8 MHz CW-doppler probe. The symmetry of arterial vessels and flow velocity in normal breasts is very high. Carcinomas appear as areas with localized flow increase. Only in two normal patients was a minimal focus of two arteries found.

Adjuvant CMFV therapy was administered from 1975 to 1984 at the Department of Gynaecology and Obstetrics of the University of Erlangen in 331 patients from a patient population, who underwent an operation with the intention of curing invasive breast cancer. Six cycles of this therapy were administered to 224 women. Only these patients were considered in the evaluation. 538 patients without axillary lymph node metastases and without adjuvant therapy served as reference population. The significance of adjuvant CMFV therapy in relation to the lymph node and menopausal status was investigated in a retrospective analysis. It was shown, that the five-year actual survivals in the group of premenopausal patients with one to three affected lymph nodes and CMFV therapy (n = 60), did not differ significantly from the group of women with premenopausally negative lymph nodes without adjuvant therapy (n = 198) (87.7% versus 88.2%). On the other hand, in the postmenopausal patients, the five-year actual survivals for the corresponding groups were 65.9% and 83.2%, respectively. The difference was significant. The results indicate, that adjuvant CMFV therapy improves the prognosis in premenopausal women, above all in cases with one to three affected lymph nodes. In the postmenopausal patients, a comparable effect could not be demonstrated in this group of patients.

Serum determinations of tumor-associated antigens in breast cancer patients are utilized for detection of metastases as well as prediction of treatment response. That is critically dependent on the validity of single daily marker determination. In our study in patients suffering from primary or metastatic breast cancer we could demonstrate fluctuations of antigens detected with the two monoclonal antibodies 115D8 and DF3 (CA-15/3) during a 24 h observation period. Beside methodical inaccuracies those fluctuating serum concentrations might be related to biological phenomena. However, on basis of our data no intra- or interindividual rhythms could be exhibited.