Prognostic significance of DNA ploidy and proliferation rate in human astrocytic gliomas

DNA ploidy and the proliferative potential in 75 gliomas were investigated using bromodeoxyuridine labelling index (BrdUrd LI), S-phase fraction (SPF) and argyrophilic nucleolar organizer regions (AgNOR) technique. There were 53 highly malignant (AIII-AIV), and 22 low-grade (AI-AII) gliomas. One fragment of the tumour was fixed in Carnoy's solution for AgNOR test, while the other fragments were used for flow cytometric determination of the labelling index, SPF and DNA ploidy. For the BrdUrdLI, tumour samples from each patient were incubated in vitro for one hour at 37 degrees C with BrdUrd using the high pressure oxygen method. The tumours showed variability in the BrdUrdLI values, SPF and AgNOR counts/cell nucleus. The same percentage of DNA aneuploidy (55%) was found in high-grade as well as in low-grade gliomas. Univariate analysis showed that patients with grade I & II gliomas had significantly higher 3-year survival rate (p = 0.0193) than those with grade III and grade IV gliomas. Also patients with lower proliferation rate of tumours (BrdUrdLI < or =2.3% and AgNOR counts < or =2.6%/cell) had higher 3-year survival rate (p<0.03), which can be helpful in prognosis. Tumour ploidy or SPF had no influence on patients' survival (p = 0.7908). Cox multivariate analysis showed that only patients' age > 45 years and high tumour grade (III and IV) were significant unfavourable prognostic factors in terms of patients' survival.     

The labelling index of human and mouse tumours assessed by bromodeoxyuridine staining in vitro and in vivo and flow cytometry

Bromodeoxyuridine (BrdUrd) incorporation and flow cytometry were used to measure human tumour kinetic parameters in vitro and in vivo. The technique was validated by comparison of labelling index estimates of mouse tumours in vivo and in vitro using BrdUrd and flow cytometry with tritiated thymidine (3HdThd) autoradiography. Similar labelling indices were obtained with both in vivo and in vitro incorporation into DNA of the two different precursors. Measurements of human tumour labelling indices were similar following in vitro incubation with either BrdUrd or 3HdThd. The use of BrdUrd allowed the visualization of a population of S-phase cells that did not appear to incorporate BrdUrd or 3HdThd. The human tumour labelling indices obtained with BrdUrd incorporation were similar to previously reported values using autoradiography studies. Preliminary studies demonstrated that significant human tumour labelling could be achieved with an intravenous injection of 500 mg BrdUrd.     

Flow cytometric DNA analysis on fine needle aspiration biopsies of liver lesions

Flow cytometric DNA analysis on fine needle aspiration biopsies of liver lesions The DNA cell content of 39 fine needle aspiration biopsies (FNAs) from five benign liver lesions, nine hepatocellular carcinomas (HCCs), and 25 metastatic tumours was analysed in a prospective fashion by flow cytometry (FCM). All benign lesions were diploid. Aneuploidy was found in five (55.6%) HCCs and in nine (36%) metastatic tumours. DNA index (DI) differences were not significant. The S-phase fraction (SPF) was higher in the malignant tumours, both combined (P < 0.02) and separated primary and metastatic (P < 0.05). We could not demonstrate an association between diploidy and percentage of benign hepatocytes in the smears of malignant tumours. The serum alpha-fetoprotein (AFP) level did not correlate with ploidy, DI, or SPF in the HCCs. In conclusion, ploidy and DI do not discriminate between benign and malignant liver lesions, but the SPF is higher in malignant tumours. DNA analysis does not help to distinguish primary from metastatic liver tumours. The presence of benign hepatocytes in samples from malignant tumours does not seem to influence the analysis of ploidy by FCM.     

Optimal conditions for immunohistochemical determination of the in vitro DNA synthesis labelling index with bromodeoxyuridine in head and neck cancer

The in vitro DNA synthesis labelling index was assessed immunohistochemically in 24 freshly obtained specimens of head and neck cancer using bromodeoxyuridine (BrdUrd) as the DNA precursor to determine the influence of BrdUrd concentration on labelling index (LI). Initially, tumour fragments were incubated in varying concentrations of BrdUrd from 2 to 100 microM for 2 h, and BrdUrd was detected with an anti-BrdUrd monoclonal antibody using immunoperoxidase labelling. There was a dose-response gradient with mean LI varying from 1.6% at 2 microM BrdUrd to 8.8% at 100 microM. The concentration-response gradient best fit a quadratic model when LI was plotted against log BrdUrd concentration (r = 0.65, P less than 0.0001). Eleven additional tumours were then studied to determine whether LI increased for BrdUrd concentrations above 100 microM. The mean LI at 125 microM and at 150 microM in these 11 tumours did not differ from the value at 100 microM, suggesting a plateau at this level. The gradient effect accounted for 17% of the variance in LI, while 60% of the variance was explained by between tumour differences. Within individual tumours, three response patterns were observed: (i) LI rose at a constant rate to the highest concentration tested (n = 8), (ii) the LI plateaued or declined at high BrdUrd concentrations (n = 6); and (iii) there was a biphasic slope slope in which the rate of rise in the LI increased at the higher BrdUrd concentrations (n = 2). The data show that BrdUrd concentration is an important variable in the immunohistochemical assessment of the in vitro LI in head and neck cancer.     

Cell kinetic studies in mouse tongue mucosa by autoradiographic, immunohistochemical and flow cytometric techniques

Abstract. A number of techniques, including autoradiography after in vivo administration of tritiated thymidine ([³H]dT), immunohistochemistry after in vivo administration of bromodeoxyuridine (BrdUrd), and flow cytometry (FCM) with and without BrdUrd detection were compared in the epithelium of ventral mouse tongue. Investigation of the diurnal proliferative rhythm by immunohistochemical detection of incorporated BrdUrd with different primary antibodies in combination with the alkaline-phosphatase-anti-alkaline-phosphatase technique, the peroxidase-anti-perox-idase method, and an indirect method with a polyclonal peroxidase-conjugated secondary antibody yielded results similar to standard autoradiography. Preparation of single cell suspensions for flow cytometry was not successful. A maximum yield of about 8.5% of the original cell number was achieved by ultrasound disintegration in combination with trypsin and dithioerythrol treatment, but neither a GdG, peak nor a G₂+ M peak was observed in DNA histograms. A better yield of about 38% of the original nuclei number was obtained by preparation of suspensions of nuclei using citric acid and the detergent Tween 20 in combination with magnetic stirring. Both S-phase index and BrdUrd labelling index could be determined by FCM and showed the normal diurnal variations. However, the BrdUrd labelling index in suspensions of nuclei was significantly higher than the labelling index determined after immunohistochemistry. The FCM S-phase index at times of day with low DNA synthesizing activity was higher than the BrdUrd index, indicating a fraction of unlabelled S-phase cells. In conclusion, detection of incorporated BrdUrd in oral mucosa by immunohistochemical techniques or flow cytometry is feasible and provides a useful tool for fast measurements of proliferation.     

DNA ploidy and S-phase fraction in carcinoma of the gallbladder related to histopathology, number of gallstones and survival.

Gallstones are a risk factor for the development of gallbladder cancer. We studied DNA ploidy and cell cycle composition by flow cytometry in archival specimens from 52 gall bladder carcinomas in relation to histopathological grade, tumour stage, gallstone number and survival. 69% of the gallbladder carcinomas showed aneuploidy. All tumours with single stones (N=11) were aneuploid while only 61% of tumours with multiple stones (N=41) were aneuploid (p=0.002). DNA aneuploidy was related to increase in T-category (p=0.01), grade (p=0.02), and nuclear pleomorphism (p=0.0005). The distribution of DNA ploidy shifted from tetraploid in low stage towards triploid positions in high stage tumours (p=0.02) combined with higher S-phase values in triploid tumours (p=0.05). S-phase fraction increased during development from normal tissue to dysplasia, cancer in situ and cancer in diploid cases (p=0.0002), and further at the change from diploid to aneuploid (p=0.004). At a median cancer specific survival time of four months patients with diploid tumours had a better survival than those with aneuploid tumours (p=0.02). In multivariate analysis of the tumour characteristic, only T-category and tumour grade were independent prognostic factors.The shift from diploid to aneuploid and the further shift of ploidy within aneuploid tumours are in agreement with the concept of a clonal development of gallbladder cancer. These changes are combined with a stepwise increase in the fraction of S-phase cells. Low frequency of symptoms in single stone patients may be the reason for detection of malignancy at a late stage of tumour development.     

S-phase fraction determined on fine needle aspirates is an independent prognostic factor in breast cancer – a multivariate study of 770 patients

Background: The prognostic significance of DNA ploidy and the S-phase fraction (SPF) have been extensively studied in breast cancer, but their clinical utility remains controversial. The type of tumour material can substantially influence flow cytometric DNA measurements. Material obtained by fine needle aspiration (FNA) biopsy is very suitable for flow cytometric DNA analysis because it contains a low proportion of non-tumour cells and less debris than tissue samples. Methods: The prognostic significance of DNA ploidy and SPF, determined on FNA samples, was analysed in 770 breast cancer patients, diagnosed between 1992 and 1997. DNA ploidy and SPF were determined at the time of diagnosis as part of the diagnostic work-up. The median follow-up was 90 months. Survival analysis included overall cancer specific survival (OS), disease free survival (DFS) and survival after recurrence (SAR). Other variables included in survival analyses were age, histological grade, histological type, lymph node status and tumour size. Disease free interval and the site of recurrence were also included in SAR analysis. Results: DNA ploidy and SPF correlated with tumour type, size, lymph node involvement and, especially, tumour grade. In a univariate analysis, both aneuploidy and high SPF were associated with shorter OS, DFS and SAR, but only SPF retained its independent prognostic significance in multivariate analyses. Independent prognostic variables for OS were node status, histological grade, SPF and tumour size. Node status, histological grade and SPF were independent predictors of DFS, while the site of recurrence, SPF, histological grade, disease free interval and age were independent predictors of SAR. Conclusions: DNA ploidy and SPF can be efficiently and routinely determined on FNA samples. High SPF is independently associated with a worse clinical outcome of patients with breast cancer. Although SPF and histological grade share prognostic information to some degree, SPF provides additional, less subjective prognostic information. The prognostic value of SPF determined on FNA samples could be even more relevant in neoadjuvant settings and for patients not amenable for surgical treatment, when histological grade cannot be assessed.     

Direct comparison of bromodeoxyuridine and Ki-67 labelling indices in human tumours

Direct comparison of bromodeoxyuridine (BrdUrd) and Ki-67 labelling indices was achieved by selecting similar areas from serial sections of human tumours. Fifteen patients were selected who had been administered BrdUrd in vivo and both proliferation markers were assessed by immunohistochemistry. The data show a good correlation between both BrdUrd LI and MIB-1 LI and T_pot (calculated using the flow cytometry derived duration of S phase) and MIB-1 LI. The contribution of BrdUrd LI to growth fraction varied as a function of proliferation characteristics. In tumours with a high LI, the number of DNA synthesizing cells represented half the growth fraction, whilst in tumours with lower LI's (<10%) the ratio of DNA precursor labelled cells as a function of growth fraction fell to between 10% and 20%. T_pot showed a linear correlation with MIB-1/BrdUrd ratio with a slope approaching unity. It was apparent that both intra- and interpatient variation in proliferation index was greater for BrdUrd labelling than for MIB-1 expression.     

Argyrophilic nucleolar organizer regions and bromodeoxyuridine and 3[H]-thymidine labelling indices in colorectal cancer

The count of argyrophilic nucleolar organizer regions (AgNORs) has been proposed as a useful method for evaluating cell replication in human tumours. The current study was undertaken to compare AgNOR values in colorectal cancers with two better established methods for investigating cell proliferation such as bromodeoxyuridine (BrdUrd) and ³[H]-thymidine (³[H]dT) labelling indices (LIs). Because some concern still exists regarding accuracy and reproducibility of AgNOR quantifying methods, we carried out a control study by independently repeating the same measurements (number, area and area per silver-stained NOR particle) in two centres with different operators and computer-assisted image analysers on 40 colorectal carcinomas. AgNOR values recorded in the two centres were strictly correlated (r= 0.75; P < 0.001 for number; r= 0.62, P < 0.01 for area; r= 0.63, P < 0.001 for area per silver-stained NOR particle) and the range of values were almost identical. Then, AgNOR values were compared with BrdUrd and ³[H]dT LIs, respectively obtained by in vivo incorporation and in vitro incubation in the same series of colorectal carcinomas. No correlation was found between AgNOR values and BrdUrd or ³[H]dT LIs. BrdUrd and ³[H]dT LIs were instead reciprocally significantly correlated. No evident correlation was seen between LIs or AgNOR values and clinico-pathological parameters of the tumour. In conclusion, in colorectal neoplasms, AgNOR values did not appear to relate with more direct parameters of cell proliferation. It follows that AgNOR reliability as a biomarker of cell proliferation remains questionable.     

Cell proliferation in human tumour xenografts: measurement using antibody labelling against bromodeoxyuridine and Ki-67

Cell proliferation was investigated in human tumour xenografts using bromodeoxyuridine (BrdUrd) labelling, evaluated either by flow cytometry or in tissue sections, and also using the proliferation marker Ki-67. BrdUrd labelling was found to increase when cryostat tumour sections were digested with an enzymic solution. This yielded a labelling index up to four times higher than that obtained using the flow cytometer. Ki-67 indices were found to be higher than those reported for human tumour biopsies, as may be expected due to the enhanced growth rate of the xenografts. Significant heterogeneity was observed in the results for cervix, breast and bladder tumours, and the results of the three methods were poorly correlated. However, three of the four tumour types showed that the tumour with the lowest Ki-67 index also had the longest potential doubling time. Since the measurement of Ki-67 index was found technically easier to perform, and also adequately reflects relative tumour cell proliferation, it is preferred over the other techniques.     

Cytokinetic analysis of lung cancer by bromodeoxyuridine labeling of cytology specimens

Recent flow cytometric (FCM) studies have indicated the prognostic value of S-phase cells (SPF) in lung cancer. More refined cytokinetic analysis can be obtained by dual-parameter FCM, labeling S-phase cells with 5-bromodeoxyuridine (BrdUrd), which can be detected using a monoclonal antibody (MoAb) to BrdUrd. Tumor cells obtained through bronchoscopic brush were incubated for 1 hr in RPMI 1640 medium with 10% fetal calf serum and 10 microM BrdUrd. After fixation in ethanol, pepsin treatment, and DNA denaturation, the nuclei were stained with anti-BrdUrd MoAb and propidium iodide. From 14 of 20 patients, sufficient material was obtained (three adenocarcinoma and seven squamous cell, one giant cell, and three small cell carcinoma). The measured SPF ranged from 5.2% to 26%. The labeling index (LI), calculated as the ratio of the number of BrdUrd-labeled cells to the total number of aneuploid cells, or diploid cells in the case of a diploid tumor, ranged from 1.2% to 16.7%; LI and SPF correlated significantly (r = 0.69). In this study, we have demonstrated the feasibility of determining the actively DNA-synthesizing cells on brush material from lung cancer cells. In addition, some extra information can be obtained about the SPF population, including the fraction of unlabeled SPF, which could be of prognostic significance.     

Intratumoral variations in DNA ploidy and s-phase fraction in human breast cancer.

To study intratumoral DNA ploidy heterogeneity and S-phase fraction (SPF) variability, we prospectively collected five different samples from 48 breast carcinomas and each sample was analysed separately by flow cytometry. Aneuploidy rate was 89.6% after analysis of four or five samples. DNA ploidy heterogeneity, i.e., different samples classified as either DNA euploid or DNA aneuploid in the same tumor was seen in 17%, and DNA index heterogeneity, i.e., tumor populations with different DNA indices (DIs) seen in different samples was 44%. A statistical model defining SPF heterogeneity is proposed. SPF heterogeneity as defined by us was 71%, and as expected the SPF heterogeneity rate increased significantly with increasing number of analysed samples. Four or more samples are needed to detect the most deviant (highest) SPF values. An unrecognized intratumor heterogeneity of DNA ploidy and SPF may partly explain the conflicting results reported in the literature on the above prognostic indicators.     

Intratumoral heterogeneity of S phase transition in solid tumours determined by bromodeoxyuridine labelling and flow cytometry

Abstract. Cell kinetics of human renal cell carcinomas xenotransplanted into nu/nu mice were analysed using the bromodeoxyuridine (BrdUrd) labelling method. Tumours were removed 0.5–14 h after injection of the BrdUrd solution. The tumour cells were stained with fluorescein isothiocyanate conjugated anti-BrdUrd antibodies and propidium iodide (DNA content). From the flow cytometry data the relative movement was calculated. Relative movement data of variable intervals after BrdUrd labelling were subjected to a fit procedure using log-normal distributions for S phase transition (T_s). The log-normal distributions were modified by inflation factors in order to get extremely asymmetric distributions. The best fits to the experimental data were obtained using wide asymmetric T_s distributions, indicating that progression through S phase in solid human tumours is considerably heterogeneous. This implies that the potential doubling time (T_pot) is longer than calculated from a single measured relative movement value obtained a few hours after BrdUrd labelling.     

Observations regarding DNA replication sites in human cells in vivo following infusions of iododeoxyuridine and bromodeoxyuridine

In studies using bromodeoxyuridine (BrdUrd) and/or iododeoxyuridine (IdUrd) to label S phase cells in cancer patients, several unique observations were made regarding DNA replication sites and the organization of newly synthesized DNA in post-mitotic cells. While the majority of tumour specimens removed at the end of infusions demonstrated concentration of replication sites around the nuclear membrane, biopsies obtained in leukaemic patients 1 week later demonstrated several distinct patterns of labelling. For example, one, two or all lobes of granulocytes were labelled. Scavenger macrophages bearing labelled leukaemic cells in their cytoplasm were also seen. Sequential IdUrd/BrdUrd labelling of solid tumours showed various patterns of nuclear/nucleolar/membrane labelling, allowing more precise localization of early versus late replication sites.     

Optimal conditions of fixation for immunohistochemical staining of proliferating cell nuclear antigen in tumour cells and its cell cycle related immunohistochemical expression

Abstract. Comparison of the results of immunohistochemical expression, such as proliferating cell nuclear antigen (PCNA) in archival material of tumours, with the clinical course is extremely valuable in determining the biological malignant potential of newly detected tumours. To obtain stable and reproducible results of immunohistochemical expression of the PCNA of tumours, we studied the optimal conditions of fNation, processing and staining of samples using animal-implanted MBT-2 cells derived from chemical-induced mouse bladder carcinoma and PC10, a monoclonal antibody for PCNA. The intensity of staining and PCNA positive rates were stable and reproducible when resected specimens from the tumours were covered with gauze wetted with physiological saline at room temperature before fixation for less than 12 h, fixation in formaldehyde was less than 48 h, and paraffin-embedded sections were dried for less than 1 h. The most clear staining of PCNA positive nuclei was observed when 10% neutral buffered formaldehyde was used as a fixative. The PCNA positive rates obtained under these conditions was compared with the bromodeoxyuridine (BrdUrd) labelling indices. Although the average PCNA positive rate was significantly higher than the BrdUrd labelling index (P?0.01), a significant correlation between PCNA positive rates and BrdUrd labelling indices was observed. In order to study the cell cycle related expression of PCNA, Ehrlich ascites tumour cells were separated by centrifugal elutriation. PCNA positive nuclei were observed in all phases of the cell cycle including G,. Occurrence of PCNA positive G₁ cells was expected at a half-life of the PCNA-protein of 20 h and a tumour cell doubling time of about 24h. Thus, the percentage of PCNA positive nuclei in a conventionally paraffin-embedded specimen of a tumour reflects both the growth fraction and the doubling time of the tumour and it may be a useful parameter of the biological malignant potential of tumours.     

Ki-67 expression and BrdUrd incorporation as markers of proliferative activity in human prostate tumour models

Abstract. The validity of the use of the monoclonal antibodies Ki-67 and anti-BrdUrd to evaluate proliferative activity of human prostate tumour models was studied. Growth of the transplantable PC-82 and PC-EW prostate tumours, as assessed by tumour volume measurements, was significantly correlated with the proliferative activity as reflected by BrdUrd incorporation into DNA ( r = 0.64 and r = 0.78, respectively). The proliferative activity of PC-82 tumours detected by Ki-67 antigen expression paralleled the pattern observed with BrdUrd ( r = 0.51) and a significant correlation ( r = 0.60) between the results obtained with both markers was found. In growing PC-82 and PC-EW tumours only small variations in the Ki-67 and BrdUrd indices were observed. In contrast, Ki-67 expression in regressing PC-82 tumours varied considerably (2.7 ± 2.2%). The BrdUrd index in regressing PC-32 tumours showed less variation (1.3 ± 0.2%), but part of the BrdUrd-positive cells were found in the stromal (murine) part of the regressing tissue. It is concluded that the Ki-67 and BrdUrd proliferation markers are reliable parameters to monitor changes in growth of prostate tumour lines, but that in slow growing or regressing tumours Ki-67 and BrdUrd data should be interpreted with caution.     

Prognostic significance of proliferative activity, DNA-ploidy, p53 and Ki-ras point mutations in colorectal liver metastases

Paired colorectal liver metastases (CLM) and normal tissue samples from a consecutive series of 36 patients were studied prospectively. MIB-1 expression was studied by immunohistochemistry on paraffin-embedded sections. DNA ploidy and S-phase fraction (SPF) measurements were performed by flow cytometry on frozen tissues. Mutations within the p53 (exons 5-8) and c- Ki-ras (codons 12 and 13) genes were detected by PCR single-strand conformation polymorphism analysis followed by sequencing. A high correlation was observed between the MIB-1 LI and SPF value (rho=0.81; P <0.01). Moreover, p53 gene mutations were associated with either high MIB-1 LI and high SPF. In univariate analysis, SPF and MIB-1 levels were related to risk of death. The association between overall survival and DNA-ploidy or p53 mutations did not reach statistical significance, but a slightly better survival was observed for patients either with DNA-diploid tumours or without mutations ( P =0.05 and P =0.06, respectively). SPF was shown by multivariate Cox model analysis to be an independent prognostic variable and thus it might be a useful prognostic factor in patients with CLM.     

Flow cytometric DNA ploidy, p53, PCNA, and c-erbB-2 protein expressions as predictors of survival in surgically resected gastric cancer patients

In order to determine retrospectively the impact of some cytometric and immunohistochemical parameters on the overall survival of gastric cancer patients treated with surgery alone, paraffin-embedded tumor samples from 137 gastric carcinoma patients undergoing curative resection from 1987-1993 were analyzed by flow cytometry (FCM) and immunohistochemistry (p53, c-erbB-2, and PCNA expression). FCM-derived parameters were DNA ploidy and fraction of S-phase cells (SPF). Multiple regression analysis was applied to determine the prognostic significance of the conventional clinicopathologic findings together with the flow cytometric and immunohistochemical parameters on overall survival. When all parameters were entered simultaneously into the Cox regression model, stage and DNA ploidy (DNA index >1.35) clearly emerged as the only independent prognostic factors. When the stages were analysed separately, the independent prognostic factors resulted DNA ploidy in early stages (I-II) and grading in stage IIIA tumors. For stage IIIB tumors, no independent prognostic factor was found. These results indicate that the DNA ploidy pattern is a valuable predictor of survival in curatively resected gastric cancer patients, especially when less advanced tumors are taken into consideration.     

Prognostic significance of proliferative activity, DNA-ploidy, p53 and Ki-ras point mutations in colorectal liver metastases

Abstract. Paired colorectal liver metastases (CLM) and normal tissue samples from a consecutive series of 36 patients were studied prospectively. MIB-1 expression was studied by immunohistochemistry on paraffin-embedded sections. DNA ploidy and S-phase fraction (SPF) measurements were performed by flow cytometry on frozen tissues. Mutations within the p53 (exons 5-8) and c-Ki-ras (codons 12 and 13) genes were detected by PCR single-strand conformation polymorphism analysis followed by sequencing. A high correlation was observed between the MIB-1 LI and SPF value (rho=0·81; P<0·01). Moreover, p53 gene mutations were associated with either high MIB-1 LI and high SPF. In univariate analysis, SPF and MIB-1 levels were related to risk of death. The association between overall survival and DNA-ploidy or p53 mutations did not reach statistical significance, but a slightly better survival was observed for patients either with DNA-diploid tumours or without mutations ( P =0·05 and P =0.06, respectively). SPF was shown by multivariate Cox model analysis to be an independent prognostic variable and thus it might be a useful prognostic factor in patients with CLM.     

In vivo labelling with halogenated pyrimidines of squamous cell carcinomas and adjacent non-involved mucosa of head and neck region

The frequency and distribution of labelled cells were studied immunohistochemically in 37 squamous cell carcinomas (SCC) of head and neck after in vivo infusion of IdUrd and BrdUrd. Tumours were classified according to their labelling patterns. Low and moderate grade SCC consisted of tumour islands separated by interstitial tissue. In some tumours labelled cells only appeared near the basal layer while in others proliferative cells were evenly distributed within the neoplastic island. In anaplastic carcinomas labelled cells were distributed either randomly or around blood vessels (cord structures). While the basal layer in adjacent normal epithelium contained very few labelled cells (LI = 1.6 ± 0.2%), the LI of basal cells in tumour islands were much higher than the average LI of the tumour (47.2 ± 2.8% and 23.8 ± 1.6%, respectively). In patients who had received cytotoxic therapy up to two months before the biopsy, the LI in the basal layer of normal epithelium was 19.0 ± 3.5%. In sequential biopsies obtained 1–2 weeks after the infusion of IdUrd and BrdUrd some labelled tumour cells were found in necrotic foci and in pearl structures. Additionally, in six tumours, we found areas of cells labelled with IdUrd alone, even though the IdUrd infusion had been followed by a BrdUrd infusion 1 h later. This is in agreement with the phenomenon of intermittent tumour blood flow described earlier in experimental tumours.