Background Several studies have employed immunohistochemistry to detect Her2/neu overexpression in urothelial carcinomas, yielding a tremendous range of positive expression rates. Few studies have examined Her2 status in non-muscle invasive bladder cancer (NMIBC) using fluorescence in situ hybridisation (FISH).
Aim To evaluate Her2 amplification in NMIBC (Ta/T1), to correlate the findings with recurrence and progression, and compare the Her2 status between primary and progressive tumours.
Methods FISH and immunohistochemistry for Her2/neu were performed on tissue arrays consisting of 36 papillary urothelial neoplasms of low malignant potential (PUNLMPs), 190 low grade urothelial carcinomas (LG-UCs) and 178 high grade urothelial carcinomas (HG-UCs). 32 cases with specimens of both primary and progressive tumours (from Ta/T1 to T2–4) were included for comparative analyses.
Results 16 HG-UCs (9.0%) showed Her2 gene amplification while none of the PUNLMPs and LG-UCs showed this aberration. There was 100% concordance in the status of Her2 amplification between primary and progressive lesions. Immunohistochemistry and FISH results were in closest agreement when overexpression was defined as 50% of tumour cells showing immunoreactivity. The cumulative incidences of recurrence and progression in Her2-amplified HG-UC were significantly higher than in those without amplification.
Conclusions A subset of high-grade NMIBCs contain Her2 amplification and are associated with markedly aggressive behaviour. Her2 diagnostics are valuable for distinguishing patients who require diligent surveillance and would potentially benefit from anti-Her2 therapies.
- Tumour Markers
- Urinary Tract Tumours
Statistics from Altmetric.com
Urothelial carcinoma is the most common neoplasm of the bladder. More than half of all bladder urothelial carcinomas present as low grade, papillary, superficial tumours, being either non-invasive (Ta) or lamina propria invasive (T1). The rest are high grade and detrusor muscle invasive.1 Low grade, non-muscle-invasive bladder cancers (NMIBCs) are usually amenable to transurethral resection (TUR) followed by intravesical instillation, while high grade, muscle-invasive tumours are eligible for cystectomy. Approximately 10–15% of NMIBCs progress to muscle invasion. Aside from histological factors such as stage and grade, it would be beneficial if other biological markers could aid in predicting the risk of progression.
The Her2/neu protein is a prognostic factor and a therapeutic target when overexpressed in breast cancer. Amplification of the Her2 gene on chromosome 17q21 is the primary mechanism for its overexpression. Her2/neu overexpression, assessed by immunohistochemistry, has been studied in urothelial carcinoma, with a wide variation in positive rates (2–80%) reported in the literature.2–15 This could be due to differences in antibodies, staining protocols and scoring systems. Fluorescence in situ hybridisation (FISH) is generally regarded as a more consistent methodology to determine the Her2 status, as DNA is better preserved in paraffin specimens and more resistant to tissue processing artefacts. There have been several investigations applying FISH to urothelial carcinoma.3 ,4 ,7 ,11–13 16–20 However, most of these have targeted muscle-invasive or metastatic urothelial carcinoma. Only two studies have focused on Her2 amplification in NMIBC.11 ,17
The aim of the present study was to evaluate Her2 gene amplification and Her2/neu protein overexpression in NMIBC. The findings were correlated to recurrence and progression. We also examined the status of Her2 amplification in paired tumour samples to see whether there is concordance between primary and progressive tumours.
Materials and methods
This study was approved by the institutional review board of Taipei Veterans General Hospital (#97-12-31A). The study subjects were retrieved from our previously published database.21 They comprised a total of 346 men and 58 women with bladder urothelial tumours who were treated by TUR from 2000 to 2002. All patients had tumours confined to the bladder. Only primary tumours were used for analysis.
The pathology specimens were reviewed and graded according to the 2004 WHO/International Society of Urological Pathology (ISUP) classification into papillary urothelial neoplasm of low malignant potential (PUNLMP), low grade urothelial carcinoma (LG-UC) and high grade urothelial carcinoma (HG-UC).1 For cases showing lamina propria invasion, care was taken to ascertain that the muscularis propria was sampled either in deep base sections or re-TUR specimens to rule out T2 disease. A total of 293 patients received intravesical instillation with mitomycin-C (N=181), epirubicin (N=50), doxorubicin (N=9), Bacille Calmette-Guérin (BCG, N=77) either as a single agent or in combination. None of the patients received anti-Her2 target therapy.
The follow-up protocol consisted mainly of periodic cystoscopy, carried out every 3 months for 3 years, followed by every 6 months for 2 years and then yearly for 5 years. Recurrence was defined as the reappearance of histopathologically confirmed urothelial neoplasm in the bladder. Cases with recurrence within 3 months of primary diagnosis were excluded, owing to the possibility of residual tumour rather than true recurrence. Progression was defined as an advance in stage from non-muscle invasive disease to detrusor muscle invasion or a more advanced stage (from Ta/T1 to T2–4), diagnosis of metastasis, or death caused by tumour (ie, muscle-invasive progression). We also calculated the all-stage progression, that is, any advance of stage (from pTa to pT1–4, from pT1 to pT2–4). Herein we prefer the muscle-invasive progression since it reflects more accurately a change of disease status that indicates aggressive treatment. As the majority of patients with T2 bladder cancer died from their disease, the statistical results of cancer-specific mortality paralleled the muscle-invasive progression (data not shown). Thirty-two cases had specimens of both primary and progressive tumours sufficient for comparative analyses.
Construction of tissue arrays
Cases with sufficient quantity were submitted to tissue arrays. Whole sections were used for small biopsies. Tissue arrays were constructed as previously described.22 We used a 16-gauge bone marrow biopsy trephine apparatus to punch the blocks. A tissue cylinder 2.0 mm in diameter was extracted. The diameter of the tissue core was significantly larger than the 0.6 mm diameter samples produced by a tissue-arrayer.
FISH was performed using a PathVysion Her2 DNA probe kit (Vysis, Downers Grove, Illinois, USA) according to the manufacturer's protocol. The kit contains a Her2 probe, labelled with SpectrumOrange, and the CEP17 probe, labelled with SpectrumGreen. The slides were examined using a Zeiss Axioplan 2 microscope (ZEISS, Göttingen, Germany) equipped with suitable single band filters. The images were acquired with a digital camera coupled with the manufacturer's software program (Micrometrics 190CU, Accu-scope, New York, USA). A minimum of 50 nuclei on each tissue spot were evaluated. Her2 was considered normal if the Her2/CEP17 ratio was ≤1.8; amplification of Her2 was indicated by a Her2/CEP17 ratio ≥2.2 or a Her2 signal visualised as uncountable clusters.23 If the ratio fell into the borderline range (1.8–2.2), another 50 nuclei were counted. We also evaluated the copy number of chromosome 17. Since the mean and SD of CEP17 count was 1.83±0.08 in 20 samples of normal bladder urothelium examined, a polysomy of chromosome 17 was defined as a CEP17 count greater than 2.07 (1.83+3×0.08).
The Hercep test (Dako, Glostrup, Denmark) was used for immunohistochemical staining. Although the recommended cut-off point for Her2/neu overexpression (3+) in breast cancer is complete membranous staining in 30% of the tumour cells,23 we were not sure whether this guideline would apply to urothelial carcinoma. Therefore, we estimated the percentage of tumour cells exhibiting complete membranous staining for each case, and correlated the results with FISH using 30%, 40% and 50% as cut-off points.
Relationships between parameters were examined using the χ2 test or Student t test as appropriate. Time to event distributions was estimated by means of cumulative incidence functions to properly take into account the patients who died (competing risk) before the end point of interest. Stepwise regression was performed for multivariate analysis. Given the presence of competing risk, it is more appropriate to plot cumulative incidence curves (Fine and Gray model) rather than disease-free Kaplan–Meier curves (Cox model).24 The status of multiplicity, tumour size, concomitant carcinoma in situ, and intravesical instillation (ie, yes or no) were entered into the multivariate model to obtain independent significance.25 Statistical tests were computed using the R-Project software.
Patient ages ranged from 23 to 99 years, with a mean of 71 years. The case numbers of PUNLMP, Ta LG-UC, T1 LG-UC, Ta HG-UC and T1 HG-UC were 36 (8.9%), 179 (44.3%), 11 (2.7%), 56 (13.9%) and 122 (30.2%), respectively. The mean follow-up times were 82 months (range 1–241 months) for patients who were alive and 36 months (range 5–138 months) for patients who died.
Table 1 lists the FISH results with respect to tumour grade and stage. None of the tumours classified as PUNLMP or LG-UC showed Her2 gene amplification, while the Her2 gene was amplified in 16 of 178 (9.0%) HG-UCs. Six (2.2%) Ta tumours and 10 (7.5%) T1 tumours showed Her2 amplification. The Her2-amplified Ta tumours were thus all high grade. Average copy numbers (mean±SD) of chromosome 17 in PUNLMP, LG-UC Ta, LG-UC T1, HG-UC Ta and HG-UC T1 were 1.97±0.54, 1.73±0.35, 1.95±0.80, 2.11±0.79 and 2.17±0.88, respectively. The frequency of polysomy 17 was significantly higher in HG-UC (p<0.0001) and T1 (p=0.01) compared to the PUNLMP/LG-UC group and Ta tumours, respectively. The mean copy number of chromosome 17 was also significantly higher in tumours classified as HG-UC and T1 compared to PUNLMP/LG-UC (2.1±0.7 vs 1.8±0.5, p<0.0001) and Ta tumours (2.0±0.7 vs 1.9±0.6, p=0.0058), respectively. The power of sample size estimation for the incidences of Her2 amplification in HG-UC and PUNLMP/LG-UC at a significance level of 0.05 was 0.985.
Correlations between FISH and immunohistochemistry
Table 2 lists the comparison between immunohistochemistry and FISH using different thresholds for defining overexpression. All Her2-amplified tumours revealed immunoreactivity in at least 40% of the tumour cells. Only one case demonstrated 40–50% positive tumour cells. If 30% was used as the cut-off as recommended for breast cancer, 14 of 30 (46.7%) immunohistochemically ‘positive’ cases were negative for Her2 amplification. The false-positive rate decreased with more stringent cut-off points. The specificity for immunohistochemistry reached 99.5% when 50% was adopted as the cut-off point, although the sensitivity decreased to 93.8%. The overall accuracy (the ratio of true assessments, ie, true positive and true negative in all assessments) was highest (99.2%) with a cut-off point of 50%. Figure 1 presents typical FISH and immunohistochemistry figures.
Association of Her2 gene amplification with recurrence and progression
The incidences of recurrence and progression in PUNLMP, LG-UC, Ta Her2-non-amplified HG-UC, T1 Her2-non-amplified HG-UC and Her2-amplified HG-UC are tabulated in table 3. In the 16 Her2-amplified HG-UCs, 11 developed local recurrence and 12 progressed within 4–52 months from the initial diagnosis (mean 23 months). The clinical features of the 16 HG-UCs with Her2 amplification are listed in table 4. Two recurrences (#348, #841) stayed at the same stage (no progression). Three patients (#224, #777, #1336) developed distant metastases without local recurrence.
Based on the aforementioned findings, NMIBC could be subgrouped into the following four prognostically distinct categories: Her2-amplified HG-UC, Her2-non-amplified HG-UC T1, Her2-non-amplified HG-UC Ta and LG-UC/PUNLMP. The cumulative incidence plots according to this stratification are shown in figure 2. Both univariate and multivariate statistical analyses demonstrated the independent prognostic significance of Her2 amplification for recurrence, muscle-invasive progression and all-stage progression after adjusting for grade, stage, status of instillation and other covariates (table 5).
We also evaluated the prognostic significance of polysomy 17. The risks for recurrence and progression in tumours with polysomy17 did not differ significantly from those without polysomy 17 within each stratum of high- and low-grade tumours (figure 3).
Comparison between primary and advanced tumours
Eight of the 32 cases with both primary and progressive tumours available for evaluation were Her2-positive in the primary lesion; all of these were also Her2-positive in the progressive tumour. The other 24 cases were negative for Her2 amplification in both primary and progressive tumours. Consequently, there was a 100% concordance of the Her2 status between primary and progressive tumours in this set.
The importance of Her2 status on the prognosis and therapy for breast cancer has been well established. However, the Her2 status regarding the choice of detection method, incidence, prognostic and therapeutic importance in urothelial carcinoma is less clear. In this series, we identified Her2 amplification in 9% of the high-grade NMIBCs examined. This frequency fell within the reported range (7–13.8%) of most prior FISH studies on invasive urothelial carcinoma.4 ,7 ,11–13 16–20 ,23 Intriguingly, Her2 amplification was not observed in any of the LG-UCs or PUNLMPs. Comparison of this finding to previous reports is difficult, given that all previous reports used the 1973 WHO classification. Previous research has revealed that among Her2-amplified urothelial carcinomas, a major proportion were grade 3 (WHO 1973) and a minor proportion were grade 2. Grade 1 Her2-amplified tumours were extremely rare. We speculate that these grade 2 tumours with Her2 amplification would likely be graded as HG-UC according to the current 2004 WHO/ISUP classification system. Our observations, taken together with those of others, suggest that Her2 amplification is an acquired oncogenetic phenomenon occurring chiefly in HG-UC, which is a genetically unstable neoplasm.
Not only did our findings demonstrate the genetic distinctness of Her2-amplified HG-UCs, but also their strikingly aggressive behaviour among NMIBCs. Her2 amplification confers a significant additional growth advantage on the tumour and distinguishes a subset of HG-UCs with a remarkably worse prognosis than those without this aberration. Although the subset of tumours with Her2 amplification accounted for less than 10% of the HG-UCs, they portended a conspicuously high risk of recurrence and progression, even when manifesting as a superficial lesion at initial presentation. Simon et al reported that Her-2 amplification was a significant prognostic factor in univariate analysis for cancer-specific mortality when urothelial carcinomas of all stages were considered.19 Similarly, Freishmann et al reported that overall survival was reduced in patients with Her2 amplification in primary urothelial carcinomas treated by cystectomy.16 However, both studies failed to confirm the independent significance of Her2 status in multivariate analyses. In this study, we recruited a sufficient number of NMIBCs to determine their risk of recurrence and progression, which are crucial issues for the management of these patients. Thus, we have verified the clinical relevance of Her2 amplification in this specific subset of tumours by using both univariate and multivariate tests.
We found that the Her2 status of the progressive disease was the same as that of the primary lesion in all matched samples. Hansel et al reported 100% concordance with respect to Her2 amplification between primary carcinomas and the paired metastases in four cases.7 Fleischmann et al similarly described high stability of Her2 status in metastases if Her2 amplification was detected in the primary tumour.16 In addition, they noted a higher frequency of Her2 amplification in nodal metastases (15.3%) than in the corresponding primary tumours (8.7%). The newly emerging Her2 amplification in the metastatic lesions is plausibly due to a late event that occurred during subsequent clonal evolution. On the contrary, the tumours studied in our series were all early and superficial, yet the Her2 status was still preserved in the initial progression of the lesions.
The agreement between Her2 FISH and immunohistochemistry has been an issue for debate. The immunohistochemical staining for Her2/neu, especially on formalin-fixed paraffin-embedded tissue, is subject to considerable technical variation. Therefore, the entire assay requires strict standardisation on each step of the procedure. We estimated immunohistochemical positivity using different cut-off points to define overexpression rather than indiscriminately using the diagnostic criteria of Her2/neu overexpression set for breast cancer to analyse bladder cancer. Our data showed a high false positive rate of Her2/neu immunostain in urothelial tumours by using the criteria (30%) for breast cancer. The accuracy was most improved with 50% positive tumour cells as the threshold. One caveat is that this percentage is not necessarily applicable to other laboratories. In order to establish the best protocol for Her2/neu immunostaining, a validation test with a set of reference samples of known Her2 status is mandatory for each laboratory. After optimal calibration, a concordance rate of 100% between FISH and immunohistochemistry has been reported.18
Given that patients with NMIBC are generally not candidates for target therapy, one may question the rationale of performing Her2/neu diagnostics. Because of the high risk of Her2-amplified HG-UC progression, the test has prognostic and therapeutic value. Patients with such tumours should be vigorously monitored. Given the preservation of Her2 alteration in the progressing tumour, patients might benefit from target therapy in case they develop muscle-invasive disease.26 Furthermore, the administration of anti-Her2/neu therapy to patients with Ta or T1 tumours that are most likely to progress to higher stages might lower the rate of disease progression; such a determination could be based on the presence of Her2 amplification. Moreover, Her2 amplification is so rare in low-grade tumours that they could be exempted from testing. We believe applying the test mainly to high-grade tumours would not impose an unreasonable burden on laboratories. For laboratories where the cost of FISH is a concern, immunohistochemistry could be used as an alternative as long as careful validation was carried out.
In contrast to Her2 amplification, polysomy 17 was not associated with increased risk for recurrence and progression. In the series of Watters et al, patients whose primary tumours contained aneusomy of 7 and/or 17 had shorter recurrence-free intervals than did patients with disomies 7 and 17 in the tumour.27 According to the report of Li et al, chromosome 17 polysomy was a prognostic indicator for overall survival.28 The apparent association of polysomy 17 with high-grade and T1 tumours might explain the prognostic effect of polysomy 17. When the tumours were stratified by grade, polysomy 17 did not provide additional prognostic information.
Because of the retrospective nature of this study, the population was mixed with respect to the scheme and duration of adjuvant therapy. On the whole, intravesical instillation has a protective effect on tumour progression.21 However, the tumours progressed in 6 of the 16 patients with Her2-amplified HG-UC despite intravesical instillation with BCG and/or mitomycin-C. The intravesical instillation status was entered into the statistical models to adjust for its effect on survival; consequently, the significance calculated from Her2 status was independent of the influence of intravesical instillation. Detailed analyses about the regimes are beyond the scope of this study. Second, the number of cases with Her2 amplification was too low to allow for further stratification, even though the data were statistically significant in certain respects. We expect to observe the same results in larger cohorts.
Her2 amplification can be identified in a subset of high-grade NMIBCs. The Her2-amplified HG-UCs are associated with a significantly heightened risk of recurrence and progression. Her2 alterations in primary tumours are highly conserved in the progressive lesions. Detection of Her2 amplification using FISH permits selection of patients with NMIBC who are at risk for tumour progression, providing the potential to benefit from anti-Her2/neu therapies.
Around 9% of high grade non-muscle invasive bladder cancer harbour Her2 amplification, while none of papillary urothelial neoplasms of low malignant potential and low grade urothelial carcinoma show this aberration.
The Her2-amplified tumours are associated with significant risks of recurrence and progression in both univariate and multivariate analyses after adjusting for grade, stage and status of intravesical instillation.
There is a high concordance in the status of Her2 amplification between primary and progressive lesions.
The authors thank the staff of the Cancer Registry Center for their kind provision of patient follow-up data.
Funding This work was supported by grant NSC98-2628-B-075-002-MY2 from the Taiwan National Science Council, Taipei.
Competing interests None.
Ethics approval This study was approved by the institutional review board of Taipei Veterans General Hospital (#97-12-31A).
Provenance and peer review Not commissioned; externally peer reviewed.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.