- Top of page
- MATERIALS AND METHODS
- CONFLICT OF INTEREST
Carcinomas of the urinary bladder are known to be related to smoking and to several chemicals, including some alkylating drugs. Several carcinogens are excreted by the kidneys and concentrated in the urine before leaving the bladder. Predominantly men are affected by bladder tumours, with a male/female ratio of 3 : 1. It is still not known to what extent bladder cancer is due to a combination of different carcinogenic factors, or how the susceptibility of the target cells might vary over time. The ageing process itself, with failing DNA repair, predisposes to bladder cancer, as the incidence increases with age [1–3]. As carcinogenic factors might operate through different genetic mechanisms [4–10], the ‘molecular genesis’ and the biological properties of the malignant cells might vary. Hence, it would be valuable to analyse comparatively tumours occurring at different incidences over time in the same population.
About 10–15 years ago, it was shown that formalin-fixed, paraffin wax-embedded tissue samples could be used for DNA extraction and mutational analysis. The same applies to immunohistochemical staining of archival tissue, in which polyclonal and monoclonal antibodies against different tissue markers are used. Such methods are also applicable to tissue specimens that have been stored for several decades. In the present study, we aimed to test whether immunohistochemical staining of biological markers, in comparison with present-day material, could be used on autopsy material from carcinomas of the urinary bladder after storage for up to 75 years. The Gade Institute has been a referral centre for autopsy and biopsy diagnostics in Western Norway since it was established in 1912. From 1930 all specimens received for biopsy diagnostics and embedded in paraffin wax have been stored systematically. Thus, tissues sampled in different periods over 75 years are available for comparative studies .
Mutations in the p53 gene and nonfunctional p53 protein in the nucleus might induce compensatory production of the protein. The mutant protein also has a longer half-life. Therefore, nuclear p53 protein accumulation is common in human malignant tumours [12,13], and p53 immunostaining was selected as the first marker in the present study. Epidermal growth factor and its receptor (EGFR) are key factors in growth regulation and often altered during carcinogenesis. Overexpression of EGFR in bladder tumours is common and correlates with the malignant potential of these neoplasms [14,15]. Therefore, we selected EGFR overexpression as a second marker. The nuclear growth suppressor p16 and two different cytokeratins were also evaluated. In addition to prolonged storage, different degrees of autolysis present in the autopsy material could destabilize the chosen markers. The aims of this study were therefore: to examine whether archival bladder cancer material from 1932–2004 could be used for immunohistochemical analyses; whether the tissue was preserved sufficiently for comparative biological analyses even on autopsy material; to establish whether the chosen markers were expressed in the same way over the time span; and presuming that the population was genetically stable could identical ‘molecular fingerprints’ be identified during the period, or whether there were different patterns.
MATERIALS AND METHODS
- Top of page
- MATERIALS AND METHODS
- CONFLICT OF INTEREST
The autopsy records were investigated for the presence of primary bladder carcinomas in the period when slides and paraffin wax blocks were stored systematically, i.e. from 1932. Material was stored at 18–20 °C in ambient air under relatively high humidity, due to the local climate. During the period the same type of fixation had been used, with submersion for several days in 4% aqueous formaldehyde solution, except that from 1985 the fixing solution had been buffered to a neutral pH. The tap water originated from the same source all the time and was ‘soft’ with a low content of calcium. Dehydration and embedding had been done manually with standardized methods in the early part of the period, while automated procedures were used from the 1960s. The mortuary where the cadavers were stored until autopsy was in the same building from 1912 to 1982, i.e. a cellar with stone walls, at a stable temperature of 5–8 °C, and from about 1940 with refrigeration to 4 °C. From 1982 the department moved into a new hospital building nearby, with a mortuary kept at 4 °C.
Based on the records, all available paraffin wax blocks were found. Most of these were in a good condition, except for some shrinkage of the paraffin wax with age. Due to brittle paraffin wax surrounding the specimen, the tissue sometimes had to be cut out manually with a knife and re-embedded into fresh paraffin wax. Thereafter the blocks were cut using a microtome, mounted on ordinary glass slides and de-paraffinized using standard procedures. From each case, one block with representative material and typical morphology was selected for immunohistochemical staining.
Autopsies with diagnosed bladder carcinomas were selected from the periods 1932–48, 1950–59, 1960–70 and 1990–2004. In addition, 13 cases with biopsies from surgically removed bladder carcinomas were selected from 1992–96 as a reference, representing optimally fixed, new material. As control to the biopsy tissue, four additional biopsies from 1932–48 and five from 1950–59 were selected. As a control of the state of the autopsy tissue, we also included 27 biopsies from the same groups of patients as the 42 autopsies in the group from 1990–2004. All the cases were re-diagnosed by a consultant pathologist (O.D.L.) on the basis of new, haematoxylin and eosin-stained slides, and typed according to the last WHO classification from 1999 . Cases of TCC with or without other types of histological differentiation (squamous and adenomatous) were collected. Scoring of a positive or negative reaction by immunohistochemical staining was only used for areas with clear TCC differentiation. In addition, the neighbouring normal urothelium and ≈ 45 samples of different normal tissues were analysed as controls.
Altogether, 144 tumours of WHO grade II and III with muscle infiltration, and some with metastases to local lymph nodes at the time of operation, were assessed (Table 1). Patient data were selected from records, including age, gender, metastasis, time from diagnosis until death, and time from death until autopsy. The study was done in accordance with the regional committee for bioethics and the Norwegian legislation on biobanks.
Table 1. The characteristics of the patients
|Period||Total with additional biopsies||Mean (sem) ||Females % of total|
|days to autopsy||age, years|
|1932–48||18||1.26 (0.17)||67.3 (2.6)||27.8|
|1950–59||30||0.94 (0.11)||66.4 (2.1)||40.0|
|1960–70||41||0.96 (0.08)||72.9 (1.4)||39.0|
|1990–04||55||1.66 (0.17)||72.1 (1.2)||34.6|
Immunohistochemistry was used on formalin-fixed and paraffin wax-embedded normal and tumour archival tissue; 5 µm thick sections were pre-treated with proteinase K (F3020, DAKO, Copenhagen, Denmark), in 0.9% NaCl, incubated with the monoclonal EGFR antibody M7239 (DAKO) diluted 1 : 25 and counterstained with haematoxylin. For detecting p53 protein the monoclonal antibody M7001 (DAKO) diluted 1 : 25 was used after pre-treatment with microwaves for 10 min at 750 W and 20 min at 500 W in 10 mm citrate buffer at pH 6. The p16 protein was stained using the monoclonal p16INK4a Ab-7 antibody (Neo Markers, LabVision, Fremont, CA, USA) diluted 1 : 100, pre-treated as previously described with citrate buffer and microwave treatment. In addition, the slides were stained with monoclonal antibodies for cytokeratin 7 (M7018, DAKO) and for high molecular weight (HMW) 34βE12 cytokeratin (M0630, DAKO). These were both diluted 1 : 50 and the sections unmasked as for EGFR protein with proteinase K in 0.9% NaCl. The staining procedures were done on the Tech Mate 500 slide-processing equipment using ChemMate Detection Kit (DAKO). This kit is based on an indirect streptavidin-biotin method where the substrate system consists of two components, i.e. a horseradish peroxidase substrate buffer containing hydrogen peroxide and concentrated diaminobenzidine solution. The substrate-chromogen produced brown staining at the site of the target antigen.
To uncover non-specific binding of the monoclonal antibodies for EGFR, p53, p16, cytokeratin 7 and HMW-cytokeratin, extra sections from each period were stained using different combinations of primary and secondary antibodies. In addition to the ordinary staining, adding both primary and secondary antibodies to the same section, a set of sections was stained with buffer in addition to the secondary antibody, and another set was stained with buffer in addition to the primary antibody.
Immunohistochemical staining of p53, p16, EGFR and the two cytokeratins was recorded as positive or negative. In addition, the proportion of positive cells in the tissue was recorded, where + denoted 1–10%, ++ 10–50%, and +++ 50–100% of the cells (not shown). Available normal tissues served as controls, both the tumour slides and slides from other organs.
The results were evaluated statistically using the chi-square test, regression analyses and in a multivariate analysis.
- Top of page
- MATERIALS AND METHODS
- CONFLICT OF INTEREST
The mean (range) interval from death to autopsy was unchanged from 1932 to 1970, at 24 (5–72) h (Table 1, Fig. 1). When all autopsies were grouped after time from death until autopsy there was no correlation between the delay and the percentage of positive immunohistochemical reactions (not shown). The morphology was well preserved in all cases, although some degree of autolysis was present (Fig. 2). As described above, several modifications were used to unmask the epitopes for EGFR, p53, p16 protein and the two cytokeratins, including proteinase K and citrate buffer with microwave treatment. Generally, strongly positive reactions were obtained at all sample times, and only a few specimens had to be re-stained due to initial weak staining (Fig. 3A–F).
Figure 2. A, Autopsy tissue from 1960 of TCC negative for (cytoplasmic) EGFR staining (×400). B Autopsy tissue from 1932 of TCC negative for nuclear p53 staining (×400). C, Autopsy tissue from 1960 of normal kidney, positive for p53 staining (unspecific) in tubuli (×200).
Download figure to PowerPoint
Figure 3. TCC from autopsies: A, haematoxylin and eosin staining in tissue from 1932 (×400), B, EGFR staining in tissue from 1965 (×100). C, p53 nuclear staining in tissue from 1932 (×200). D, p16 staining in tissue from 1960 (×100). E, Cytokeratin 7 staining in tissue from 1943 (× 100). F, HMW-cytokeratin staining in tissue from 1960 (×200).
Download figure to PowerPoint
As a first step, all paraffin wax-embedded tissue from normal organs was tested and found largely negative for EGFR. This included connective and fatty tissue, tissue from thyroid gland, smooth muscular, leukocytes, heart, lungs, uterine cervix, normal urinary bladder epithelium, appendix, lymph nodes, endometrium, ureter and ovaries. In five samples of liver tissue, four showed weakly to moderately positive hepatocytes. One sample of brain tissue showed positive staining in some of the glial cells, and the same was noted for some of the cells in bone marrow. In one case with a bladder carcinoma infiltrating near to the prostate, there was strongly positive EGFR staining in the glandular epithelium. The other cases were only weakly positive. In one case of bladder carcinoma, EGFR reactivity was predominant in the normal urothelium.
For p53 there was a negative reaction (<1% positive stained nuclei) in all the examined normal tissues. For p16 and the two cytokeratins there were strongly positive reactions. P16 was positive in the cell nuclei of all tested organs. Cytokeratin 7 showed positive staining in bile ducts, bladder mucosa, kidney and bronchial epithelium at all sample times, and negative staining in mesenchymal tissues. The reaction for HMW-cytokeratin was positive in normal liver parenchyma, bladder and urethral mucosa, and in the prostate, thyroid gland and epithelium of the seminal vesicle (Table 2).
Table 2. Immunohistochemical marker reactions in available normal tissue from autopsies 1932–65 (0, not available)
|Tissues||No. of samples||HMW- cytokeratin||Cytokeratin 7||p16|
|Thyroid gland|| 1||−||−||−|
|Liver, parenchyma|| 5||−||−||0|
| bile ducts|| ||−||+||0|
|Urinary bladder epithelium|| 2||+||0||0|
|Bone marrow|| 1||−||−||0|
|Connective tissue|| 6||−||−||−|
With differential counting there were >10–20% strongly positive cells in all the bladder tumours that stained positively for one or more markers. In two-thirds of the cases 50–80% of the cells were strongly positive. There was a clear demarcation between positive and negative cells in all tumours. The negative tumours showed no staining, as did the tissue of normal neighbouring organs (Fig. 3A–F).
Cytokeratin 7 positivity increased by 13% in the 70 years, but cross-tabulation analyses showed no significant trend across time (years in groups) for this cytokeratin (P = 0.32; Fig. 4D).
Figure 4. The percentage positive score in the tumour tissue of cases from 1932 to 2004 of A HMW-cytokeratin; B, p53; C, p16; D, cytokeratin 7; and E, EGFR.
Download figure to PowerPoint
About 90% of the tumour cells showed positive staining for HMW-cytokeratin in tumours from 1932–48. The proportion of positive cases gradually decreased to ≈ 30% in 2004, with a significant trend over time (cross tabulation analyses; P < 0.01; Fig. 4A). Logistic regression was further used to estimate the strength of this association, and the odds of a positive HMW-cytokeratin score in a tumour decreased significantly with time (odds ratio 0.352; P < 0.01). The presence of metastasis was also significantly correlated with a positive score for HMW-cytokeratin (odds ratio 3.161; P = 0.02). The influence of the two fixed variables (years in groups and metastasis) on each other’s effect on HMW-cytokeratin score was estimated in a two-block logistic regression analysis; this did not significantly modify the correlations found initially.
For p53 immunoreactivity, there were higher fractions of positive cells with time; in tumours from 1990 to 2004, the number of positive cases increased by ≈ 20% from 1932 to 1948. Cross tabulation analyses showed a non-significant trend with time for p53 (P = 0.06; Fig. 4B). The positive cases showed strongly positive reactions in the nuclei, in at least 10–20% of positive cells, and often more. Thus, the staining procedures seemed to give optimum results when positive (Fig. 3C). There was no correlation between EGFR and p53.
The p16 positive tumours showed no significant variations with time. Most positive cases were in 1990–2004 and fewest in the two middle decades (1950–70; P = 0.14; Fig. 4C). There was a significant correlation between the p16 score and the grade of the tumour (P = 0.04) by cross tabulation analysis, but the strength of the association was at the borderline of significance (P = 0.055, by logistic regression analysis). The odds of a positive score for p16 decreased with higher grade (odds ratio 0.461).
The mean percentage of EGFR-positive cases was 74% over all four periods; the percentage of positive cases declined gradually during the periods, the most recent biopsies/autopsies having a lower, but not significantly so, proportion of positive tumours than the earlier samples (P = 0.21; Fig. 4E). Cross tabulation analysis between the other markers, regardless of time, showed positive correlations between EGFR and HMW-cytokeratin (P = 0.008) and between p16 and cytokeratin 7 (P = 0.009).
All data were also evaluated in a multivariate analysis (general linear model) in which years in groups, age in groups of 10 years, gender, tumour grade, the presence of metastasis, time from diagnosis to death (months), and time from death to autopsy (days) were included as fixed factors. The analysis showed that gender, age, time from diagnosis to death and time from death to autopsy had no significant influence on the dependent variables (all the protein markers used).
Comparing positive score for biopsies and autopsies from the same patients in the group from 1990–2001 showed a concordance of >95% for the two types of specimen. Autopsy material showed no systematic underscoring. For EGFR, cytokeratin 7 and HMW-cytokeratin the positive score were similar in the autopsy and biopsy material, while p53 and p16 had slightly higher scores in the biopsy tissue (Table 3).
Table 3. Positive score in 27 autopsies and biopsies from the same patients in the period 1990–2004
|HMW-cytokeratin|| 5/22|| 4/23|
The control sections stained with the monoclonal antibodies for EGFR, p16 and the two cytokeratins were negative for both combinations of staining (buffer in addition to secondary antibody, and buffer in addition to primary antibody; Fig. 2A). Control staining with the monoclonal antibody for p53 was also negative except for sections of normal kidney tissue that had a positive cytoplasmic staining of the tubuli using both combinations (Fig. 2B,C).
In addition to the present material comprising only TCC, we collected four more cases from the 1930s with squamous cell carcinomas. When these were added to the total material and matched to seven other cases of squamous cell carcinoma from 1950 to 2004, the proportion of p53-positive cases was significantly higher in the most recent group than in 1930s (P < 0.05; data not shown).
- Top of page
- MATERIALS AND METHODS
- CONFLICT OF INTEREST
The immunohistochemical detection of nuclear and cytoplasmic markers (EGFR, p53, p16, cytokeratin 7 and HMW-cytokeratin) was as good in old as in new paraffin-embedded material (Table 2). In such archival tissue, autolysis after death, prolonged storage and lack of standardization of tissue fixation could theoretically impair the results of histochemical and molecular analyses. However, the present tested material gave optimum histological sections and immunohistochemical staining. That the time from death to autopsy was rather short might be part of the explanation for the successful analyses. On average, the autopsy was within 24 h after death in the period 1932–70; this delay increased to 1.7 days in the 1990s (Fig. 1; Table 1), a development coinciding with the decreasing rate of autopsy in our hospital.
Interestingly, no tumour cases showed weak staining only, as there was either strong or entirely negative staining with high specificity. The normal control tissues gave similar results, and there were distinctly positive reactions in normal liver and brain. This indicates that the epitopes of the used markers are well preserved in archival specimens. However, successful staining depended on appropriate unmasking of the epitopes with proteinase K and citrate buffer (pH 6). It cannot be excluded that the relevant epitopes are degraded or chemically modified in the negative tumour specimens. This is also well known with immunohistochemistry on freshly embedded tissue, but if this had been a problem it would be expected that old tumour specimens showed weak or faded positive staining; such patterns were not apparent (Fig. 3A–F). This enabled a comparative study on specimens stored from the early 1930s until the present and allows some conclusions to be drawn from the distribution of positive and negative cases (Fig. 4A–E).
While staining for EGFR, cytokeratin 7 and p16 were roughly constant over the period of 70 years (no significant differences), the staining patterns of p53 and HMW-cytokeratin varied. Over the decades, the proportion of p53- positive tumours moderately increased (borderline significance), while HMW-cytokeratin correspondingly decreased to a third; the latter is generally a marker of keratinizing basal cells. In addition, Varma et al. reported that HMW-cytokeratin is a sensitive marker for high-grade invasive urothelial carcinomas originating in the prostate. As this cellular component was more common in bladder carcinomas in the 1930s than 70 years later, it could indicate a gradual change to a less malignant phenotype.
By contrast, the finding that nuclear p53 accumulation was more dominant in recent times might suggest a higher malignant potential . However, the possibility that p53 protein becomes more degraded with time in archival tissue, and is thus rarer in the oldest material, cannot be excluded. For comparison, glioblastomas in the brain have two main subtypes; one develops through progression from a low-grade astrocytoma, mainly occurs in the young and is termed secondary glioblastoma. Early mutations of TP53, and subsequently nuclear accumulation of p53 protein, dominate in this cancer type. The other tumour type arises de novo in older patients; EGFR overexpression dominates, and TP53 mutations are rare in these lesions [19–26]. Possibly, TCC might represent a parallel case.
The mean age of the patients in the four groups was ≈ 65–70 years, and differences between the variables due to variation in age are thus excluded (Table 1). In addition, the tumours were biologically and clinically comparable. As seen in Table 1, the age of the patients at death increased significantly after 1960. During the same time there was an increase in the incidence of bladder cancer . The increased lifespan of patients with bladder cancer could possibly be due to improved treatment, but as indicated above, also to a less malignant phenotype since 1960.
As both p53 and EGFR overexpression are accompanied by a negative prognosis, they might represent different ‘molecular fingerprints’ in high-grade bladder carcinomas in the four periods. Interestingly cross tabulation analysis between the markers, regardless of time, showed a significant correlation between EGFR and HMW-cytokeratin, and with p16 and cytokeratin 7. In addition, the multivariate analysis showed a significant correlation between p16 score and the grade of the tumours. The odds of a positive score decreased with higher grade. This confirms previous suggestions that p16 accumulation is involved in low-grade and early-stage bladder cancer, and that expression of this protein decreased with increasing grade [27,28]. The multivariate analysis also showed a significant correlation between the presence of metastasis and a positive score for HMW-cytokeratin.
Smoking is considered a major causal factor for human bladder carcinomas; as it might take 10–20 years or more of smoking for such a tumour to develop, it might be difficult to assess the impact of this factor in present material. The clear preponderance of males in the present samples is compatible with the smoking habits in Norway during the 1930s and 10–20 years ago. At that time only a few women were daily smokers. However, the proportion of daily smokers among men was even higher in the 1930s than in 1980–90 . Differences in smoking habits can therefore not explain the present differences in biological markers as shown by immunohistochemistry. Possibly, other environmental factors might contribute to bladder carcinogenesis and the increased incidence of bladder tumours during the 20th century. Invasive bladder carcinomas are classified by others as having high or low levels of chromosomal aberrations . At least two different molecular pathways for initiation and development of bladder cancer are known to exist. From our present pilot study it might be hypothesized that bladder carcinomas developed through different molecular pathways in the 1930s and in the 1990s; the 1960s seem to represent an intermediate period.
Taken together, there seems to be a time-dependent pattern of biological features in TCC; in the 1930s these tumours tended to have a high proportion of HMW-cytokeratin- and EGFR-positive cases, combined with more metastases and a shorter patient lifespan. Seventy years later there was a tendency for the opposite pattern. The finding during this period of an inverse relation between HMW-cytokeratin and EGFR overexpression on the one hand, and nuclear p53 protein accumulation on the other, might indicate that the development of bladder tumours follows different molecular genetic pathways in the periods compared. Theoretically, this could correspond to exposure to different environmental and carcinogenic influences during the periods.
Retrospective biological diagnostics can serve as a valuable tool for retrieving rare types of tumours and for monitoring malignant disorders over long periods, back to times with an environment totally different from the present. A huge resource of material in tissue samples is stored as paraffin wax blocks in pathological departments around the world; in Scandinavia alone there are ≈ 100 million blocks and about twice as many slides. These represent various types of tissues and diseases over 75 years, and originate in patients of different ages. DNA and RNA are degraded, but fragments can be extracted and used for studies on mutations, deletions and overexpression, and to some extent also for gene expression studies .
We conclude that with proper handling and modern improved methods, paraffin wax-embedded autopsy material stained with immunohistochemical markers can be compared over 70 years. The approach strongly suggests that carcinomas in Western Norway have altered their phenotypic expression over this time span.