The role of fluorescence in situ hybridization to predict patient response to intravesical Bacillus Calmette-Guérin therapy for bladder cancer: A diagnostic meta-analysis and systematic review.

BACKGROUND: The aim of the study was to systematically review the relevant studies to assess the role of fluorescence in situ hybridization (FISH) test for predicting patient response to Bacillus Calmette-Guérin (BCG) therapy after transurethral resection of bladder tumor (TURBT).
METHODS: We searched PubMed, Embase, and the Cochrane Library from inception to July 5, 2018, and used Quality Assessment Tool for Diagnosis Accuracy Studies (QUADAS-2) to assess the quality. We pooled sensitivity, specificity, and area under curve (AUC) of baseline and post-BCG FISH test for predicting tumor recurrence. Hazard ratio (HR) with 95% confidence intervals (95% CIs) and a Fagan nomogram were applied to assess predictive accuracy of post-BCG FISH test.
RESULTS: A total of 6 studies with 442 participants for post-BCG test and 404 participants for baseline BCG test were included. The pooled analysis for post-BCG FISH test revealed the sensitivity of 0.54 (95% CI 0.38-0.69), specificity of 0.84 (95% CI: 0.72-0.91), and area under the curve (AUC) of 0.78 (95% CI: 0.74-0.81) for predicting tumor recurrence. Patients with positive post-BCG FISH test were more likely to recur during follow-up (HR 3.95, 95% CI 2.72-5.72). The Fagan nomogram revealed the "post-test" probability of tumor recurrence increased by 29% for patients with positive post-BCG FISH test. The baseline FISH test had a pooled sensitivity of 0.70 (95% CI 0.55-0.81), specificity of 0.41 (95% CI: 0.26-0.58), and AUC of 0.60 (95% CI: 0.56-0.64) for predicting recurrence.
CONCLUSION: The post-BCG FISH test can predict BCG failure with high specificity and patients with positive post-BCG FISH test were more likely to recur. However, the relatively low sensitivity of post-BCG FISH test and unsatisfactory performance of baseline FISH test may limit their mono-use.

Introduction

Nonmuscle-invasive bladder cancer (NMIBC) is a group of superficial tumors of the bladder (stage Ta, T1, and carcinoma in situ), which accounts for 70% to 75% newly diagnosed bladder cancer.[ Depending on prognostic factors, patients are categorized into low-, intermediate-, and high-risk groups to predict tumor recurrence and progression.[ The most effective treatment for intermediate- or high-risk NMIBC is transurethral resection of bladder tumor (TURBT) combined with intravesical instillation of Bacillus Calmette–Guérin (BCG).[ Nevertheless, as high as 42% of patients developed BCG failure in 1 year and a large portion of these tumors recur or progress into muscle-invasive disease despite the combined therapy.[

For patients not responding to intravesical BCG, accurate prediction BCG treatment failure is critical to help decide early cystectomy as well as avoid overtreatment.[ Conventional cytology can detect high-grade urothelial carcinomas with high diagnostic accuracy. However, it has difficulty distinguishing inflammatory and reactive changes from the recurrent tumor, especially in patients being treated with intravesical BCG.[ The cystoscopy during follow-up is an important prognostic indicator.[ But cystoscopy mainly depends on subjective visible changes and relies on detection of actual tumor recurrence, which made it also a poor predictor for early intravesical BCG therapy failure.[

The UroVysion FISH test uses fluorescently labeled DNA probes to identify chromosomal alterations.[ As the chromosomal integrity is not affected by hematuria, infections of urinary tract, or any other instrumentation process, FISH is reported to have better accuracy and earlier detection of tumor recurrence than cytology or cystoscopy.[ A recent update of the American Urological Association (AUA) guideline mentioned that the UroVysion FISH test may be used to assess response to intravesical BCG therapy.[ However, the evidence was limited to expert opinion. It remains unclear whether a review of the current literature would lead to a revision of clinical guidance. Therefore, we performed a diagnostic meta-analysis to assess the predictive accuracy of the UroVysion FISH test for predicting tumor recurrence in bladder cancer patients receiving regular intravesical BCG after TURBT.

Materials and methods

Search strategy

The systematic search was performed according to the PRISMA statement.[ PubMed (Medline), Embase database, and the Cochrane Library were searched up to July 5, 2018. We also manually screened the references of included studies for additional citations. The search strategy was applied to identify all trials by using medical subject headings terms in combination with keywords of urinary bladder neoplasms, bladder cancer, BCG vaccine, fluorescence in situ hybridization, and FISH. We limited studies to human.

Selection criteria

Two authors (Y.B and X.T.) evaluated all potential articles independently. The inclusion criteria included a cross-sectional study that assessed the accuracy of FISH test (baseline FISH test: the specimens of urine were collected before starting the first BCG instillation; and post-BCG FISH test: the urine samples were collected when finishing the 6-week course of BCG instillation) for predicting BCG response. Positive FISH test was defined that at least 4 cells showed polysomy on at least 2 chromosomes (3, 7, or 17) and/or there were at least 12 cells with no signal (homozygous deletion) for 9p21.[ Recurrence/progression status were used as endpoints. Tumor recurrence was defined as biopsy-proven NMIBC during the follow-up time, and tumor progression was defined as the development of muscle-invasive disease. Reporting data were available to calculate the true-positive, true-negative, false-positive, and false-negative rates of the FISH test. Review articles, editorial comments, conference reports, and low-quality studies were excluded. Any discrepancies were resolved by a third reviewer (Q.W.).

Data extraction

Two authors (Y.B. and X.T.) performed data extraction independently. We extracted data, including the first author, publication year, country, study design, participant details (number of patients, age, sex, and pre-instillation parameters), FISH test details (FISH system, sample number, and collecting time), follow-up time, and outcome data (recurrence and progression). Outcome data of predictive results detected by baseline and post-BCG FISH test were extracted in 2 × 2 contingency tables.

Evidence quality assessment

The Quality Assessment Tool for Diagnosis Accuracy Studies (QUADAS-2) was used by 2 authors (Y.B. and X.T.) separately to evaluate the quality of each trial, which included 4 domains: patient selection; index test conduct; reference test conduct; and participant flow and timing.[ We judged a study to have “low risk of bias” if it was evaluated as “low” on all 4 domains or the first 3 terms concerning applicability. A study might be evaluated as a high risk of bias if more than 1 domain (including one) was judged “high” or “unclear.” Any discrepancies were resolved by a third reviewer (Q.W.).

Statistical analysis

The pooled sensitivity, specificity, and the summary receiver operating characteristic (SROC) curve with an area under the curve (AUC) of baseline FISH test and post-BCG FISH test for predicting tumor recurrence were calculated using the STATA, version 12.0 (Stata Corporation, College Station, TX), respectively. The hazard ratio (HR) with 95% confidence intervals (95% CIs) were synthesized to compare the prediction capacity of tumor recurrence for positive and negative post-BCG FISH test. A Fagan nomogram was used to assess the predicting value of the post-BCG FISH test for tumor recurrence using likelihood ratios to calculate a post-test probability based on Bayesian theorems. Heterogeneity was valued with the Chi-square and the Higgins–Thompson I2 method. The pronounced heterogeneity was indicated by a P value < .05 and an I2 > 50%.[

Results

Characteristics of the included studies and participants

The initial search identified 151 records in total. We identified 11 records for full-text review after screening titles and abstracts. Among these records, 4 studies were reviews, and 1 study was a duplicate. Finally, data from 6 studies[ were synthesized in this diagnostic meta-analysis (Fig. 1). We summarized the patient clinical features and essential data in Table 1.

Figure 1

Flow diagram illustrating numbers of studies in the meta-analysis.

Table 1

Characteristics of the included studies and participants.

Predictive accuracy of post-BCG FISH test for tumor recurrence

Overall, 442 participants received the post-BCG FISH test with 31.9% (141/442) resulting in FISH positive and 37.3% (165/442) developing tumor recurrence during the follow-up time. The statistical analysis revealed the pooled sensitivity of 0.54 (95% CI 0.38–0.69) and pooled specificity of 0.84 (95% CI: 0.72–0.91) (Fig. 2). An overall accuracy was revealed by the SROC curve with AUC of 0.78 (95% CI: 0.74–0.81) (S. Figure 1). Patients with a positive post-BCG FISH test were more likely to develop recurrence than patients of the negative post-BCG FISH result (HR 3.95, 95% CI 2.72–5.72). There was no heterogeneity identified (I = 0.00, P = .552) (Fig. 3). The Fagan nomogram illustrated that with a positive post-BCG FISH result, there was a 66% “post-test” probability of a subsequent tumor recurrence episode, and the “post-test” probability of tumor recurrence dropped to 24% while with a negative post-BCG FISH test (Fig. 4).

Figure 2

Forest plots of sensitivity and specificity for post-BCG FISH test predicting tumor recurrence.

Figure 3

Forest plot of hazard ratio (HR) of tumor recurrence for positive and negative post-BCG FISH test. CI = confidence interval.

Figure 4

Fagan nomogram of post-BCG FISH for predicting tumor recurrence. With a positive post-BCG FISH result, there was a 66% “post-test” probability of a subsequent tumor recurrence episode, and the “post-test” probability of tumor recurrence dropped to 24% while with a negative post-BCG FISH test. LR = likelihood ratio.

Predictive accuracy of baseline FISH test for tumor recurrence

Of the 404 participants examined by the baseline FISH test, 60.9% (246/404) resulted in FISH positive, and 36.1% (146/404) developed recurrence. The sensitivities were between 0.44 and 0.79, while the specificities were between 0.12 and 0.70. The statistical analysis revealed the pooled sensitivity of 0.70 (95% CI 0.55–0.81) and pooled specificity of 0.41 (95% CI: 0.26–0.58) (Fig. 5). An overall accuracy was revealed by the SROC curve with the AUC of 0.60 (95% CI: 0.56–0.64) (S. Figure 2).

Figure 5

Forest plots of sensitivity and specificity for baseline FISH test predicting tumor recurrence.

Predictive accuracy of combined baseline and post-BCG FISH test for tumor recurrence

Three studies[ have investigated the predictive capacity of tumor recurrence when regarding the combination of baseline and post-BCG FISH test (Table 2). Overall, simultaneously, positive results for both baseline and post-BCG test predicted a higher risk to recur. When regarding the impact of combined tests on recurrence rate, all 3 studies observed the lowest recurrence rate in patients with simultaneously negative tests, while the portion of recurrence cases remained highest in patients with negative baseline FISH result but transforming to positive after receiving instillation of BCG compared with other 5 groups.

Table 2

Predictive accuracy of combined baseline and post-BCG FISH test for tumor recurrence.

Predictive accuracy of baseline and post-BCG FISH test for tumor progression

Three studies[ have reported the predictive capacity of baseline and post-BCG FISH test for tumor progression during follow-up. However, a pooled analysis was not available due to the limited number of studies. We therefore extracted necessary data for 2 × 2 table and calculated the sensitivities and specificities separately. All studies reported that post-BCG FISH can predict tumor progression with both sensitivity and specificity above 0.50. For baseline FISH, the specificities varied; however, all 3 studies revealed that the sensitivity can reach above 0.50 (S. Table 1).

Quality assessment of the included studies

Quality evaluation of individual studies is summarized in Table 3. In the patient selection domain, 1 study[ was considered to be at a high risk for its retrospective design. For the flow and timing domain, we considered 2 studies[ at a high risk of bias for existing withdrawals during the follow-up period. All of the other domains were considered to be at a low or unclear risk.

Table 3

Quality assessment of the included studies.

Evaluation of heterogeneity and sensitivity analysis

There was substantial heterogeneity between studies. Although post-BCG FISH test was defined that the urine samples for test were collected immediately after finishing the 6-week course of BCG instillation in most studies. In 2 studies,[ the urine samples were collected a period of time after the last instillation of BCG at sixth week. The sensitivity analysis by removing the 2 studies revealed that the pooled sensitivity of 0.53 (95% CI 0.30–0.74) and specificity of 0.86 (95% CI, 0.70–0.94), which indicated the short interval after BCG therapy may not have much impact on the final results (S. Figure 3). Similarly, sensitivity analysis by removing the study by Savic et al[ in which a small portion of urine specimens of baseline FISH collected before TURBT revealed consistent results for baseline FISH test with a sensitivity of 0.69 (95% CI 0.51–0.82) and a specificity of 0.42 (95% CI 0.22–0.65) (S. Figure 4). Sensitivity analysis (data not shown) by removing the studies[ of high risk revealed pooled specificity of 0.80 (95% CI 0.68–0.88) for post-BCG test, pooled sensitivity of 0.71 (95% CI 0.54–0.84) for baseline FISH test, and a similar trend to recur for patients with positive post-BCG FISH test (HR 3.80, 95% CI 1.98–5.61). Because there were less than 10 studies available, we did not assess the publication bias.

Discussion

In this diagnostic meta-analysis, we found that the post-BCG FISH test can predict tumor recurrence with high specificity and acceptable sensitivity. Patients with a positive post-BCG FISH were more likely to develop recurrence than patients of the negative post-BCG FISH test.

Intravesical BCG instillation is well known for its antitumor effect in bladder cancer. However, BCG failure does happen, and patients recur or progress. Attempts have been made to identify good predictors such as surveillance cystoscopy and cytology for tumor recurrence during intravesical BCG treatment.[ However, due to the fact reactive urothelial changes could compromise the accuracy of cytology or cystoscopy,[ currently, no diagnostic tool is available to discriminate between BCG responders and BCG failures. The FISH test combines the morphological changes of conventional cytology with molecular DNA alterations and was reported to have better performance than cytology in the detection of noninvasive papillary tumors in the bladder.[ Considering that the integrity of chromosome is not affected by BCG, the FISH test could be a candidate tool to predict BCG failure and predict tumor recurrence.

In our study, the relatively higher specificity revealed the post-BCG FISH test seemed a better predictor for assessing response to intravesical BCG therapy compared with baseline FISH test. The difference was hypothesized that patients did not have benefitted from BCG induction therapy yet at baseline time.[ And multiple studies[ have reported the persistence of genetically aberrant cells after TURBT, which necessitated adjuvant therapy but not eventually progress to recurrence. Meanwhile, as the median time to recurrence after BCG reported was reported to be 26 months,[ the median follow-up time of 16 to 23.4 months in this pooled study may not be long enough to detect all recurrences, which may also contribute to the false-positive rates.

The relatively low sensitivity of the post-BCG FISH test meant that there would still be nearly 46% negative patients with probability of recurrence, but still better than cytology.[ And it was interesting to observe that post-BCG FISH test would have a lower sensitivity than pre-BCG FISH. On the one hand, the restriction to 4 chromosomes (3, 7, 9, 17) in FISH test, which is only part of chromosomal alterations in patients with bladder tumors, may contribute to the low sensitivity.[ On the other hand, FISH test relies on sufficient malignant cells in the urine sample. The aggressive exfoliation of the vesical mucosa instead of tumor cells during BCG therapy could preclude the fulfillment of the positive criteria, which may lower the sensitivity of post-BCG FISH test.[ Anyhow, previous studies reported the false-negative results mostly appeared in low-grade (grade 1 and 2) bladder tumor patients because these tumors had relatively few, if any, chromosomal abnormalities.[ It might not to be a major concern because the 5-year progression rate is only 2% to 4% for these low-grade tumors.[

Considering the better predictive role when combing baseline and post-BCG FISH test together, multiple FISH tests that were longitudinally taken during a patient's follow-up may be of more value. For example, a patient with negative baseline FISH test but transforming to positive after receiving instillation of BCG recur easily even when compared with a patient with simultaneously positive FISH tests. This indicates that doctors should pay more attention to those patients with inconsistent FISH tests and provide necessary interventions especially for those transformed from negative to positive. However, no sufficient data can be pooled for synthesizing in current studies. More studies are needed to figure out the true role of combined FISH tests. And as tumor progression is more precise to identify patients who are going to need a cystectomy, more studies assessing the role of FISH test in predicting tumor progression are needed in terms of better implication for practice.

As far as we know, this was the first meta-analysis to assess the predictive accuracy of the UroVysion FISH test in predicting response to intravesical BCG therapy in patients with NMIBC. Nevertheless, limitations existed in our study, with the most obvious one being the substantial heterogeneity in varied aspects. However, the sensitivity analysis revealed that the minor interval for FISH test had a little impact on the final results. And as the diagnostic property of the study and FISH test depends mainly on the chromosomal alterations, varied criteria for intravesical BCG therapy, the different protocol for intravesical therapy as well as tumor prevalence may have a little impact on the final results. Second, the limited number of studies available to synthesize was also one major limitation of our study. However, we have conducted a comprehensive search and included all studies available to synthesize the results.

In summary, we analyzed the current clinical evidence to assess the predictive accuracy of the UroVysion FISH test for predicting BCG response in patients with NMIBC. We found that the post-BCG FISH test can predict tumor recurrence with high specificity and patients with a positive post-BCG FISH were 3.95 times more likely to develop recurrence. However, the relatively low sensitivity of post-BCG FISH test may limit its utility. The combination of post-BCG FISH test with cystoscopy/cytology as well as baseline FISH test may be a promising method to help detect bladder cancer recurrence as early as possible. And more studies assessing the role of FISH test in predicting tumor progression are needed in terms of better implication for practice.

Acknowledgments

We thank Ian Charles Tobias (Western University of Ontario) for reviewing the manuscript.

Author contributions

Conceptualization: Yige Bao, Qiang Wei.

Data curation: Xiang Tu, Tiancong Chang, Shi Qiu, Lu Yang, Jiwen Geng, Liuliu Quan.

Formal analysis: Yige Bao, Xiang Tu, Shi Qiu, Liuliu Quan.

Funding acquisition: Yige Bao, Qiang Wei.

Methodology: Yige Bao, Xiang Tu, Shi Qiu.

Software: Tiancong Chang, Lu Yang, Jiwen Geng.

Supervision: Qiang Wei.

Writing – original draft: Yige Bao, Xiang Tu.

Writing – review & editing: Yige Bao, Xiang Tu, Tiancong Chang, Shi Qiu, Lu Yang, Jiwen Geng, Liuliu Quan, Qiang Wei.