Bladder preservation approach versus radical cystectomy for high-grade non-muscle-invasive bladder cancer: a meta-analysis of cohort studies

Bladder cancer (BC) has a high morbidity in patients worldwide. As the 9th most commonly diagnosed cancer and the 13th most common cause of death worldwide, BC caused 188,000 deaths in 2015 worldwide [1, 2].

Among the 81,190 estimated newly diagnosed BC in the US in 2018, nearly 75% were non-muscle-invasive bladder cancer (NMIBC). NMIBC is defined as a superficial neoplasia confined to the mucosa, (including Ta which is a noninvasive papillary carcinoma and carcinoma in situ (CIS) which is flat and non-papillary) or lamina propria (T1) based on the American Joint Committee on Cancer (AJCC) staging system, also known as the tumor node metastases (TNM) classification [3‐5]. Histologically, BC is generally graded using the 1973 World Health Organization (WHO) classification system or the 2004 revision. The 1973 version comprises grade 1 to 3 and is based on the degree of cellular anaplasia compatible for the diagnosis of malignancy. Grade 1 (G1) applies to tumors having the least degree, grade 3 (G3) applies to tumors having the most severe degree, and grade 2 (G2) lies in between. The 2004 revision categorizes tumors into “low-grade (LG)” or “high-grade (HG)” depending on the neoplasm of the urothelium lining papillary fronds. This may show either an orderly appearance or with easily recognizable variations in architecture and cytologic features or a predominant pattern of disorder with moderate-to-marked architectural and cytologic atypia [6, 7]. Both grading systems have confirmed prognostic value and have been accepted widely. However, which one is more clinically significant remains controversial [8‐10]. In NMIBC, high-grade non-muscle-invasive bladder cancer (HGNMIBC) has the highest risk due to its aggressive clinical, biologic, and histopathologic characteristics [11, 12]. A 15-year study demonstrated that high-grade T1 bladder cancer had a 50% progression rate and had a mortality of 30% during patient follow-up [13]. Hence, the management strategy for HGNMIBC is an unmet clinical need.

Regarding bladder preservation (BP), intravesical Bacillus Calmette–Guerin (BCG) following transurethral resection (TUR) has been the gold standard for over 40 years and has been demonstrated to decrease recurrence rates, progression rates, and mortality in high-risk NMIBC patients [14‐16]. However, several studies have reported that 23 to 74% high-grade T1 tumor recurred and more than 50% progressed after receiving intravesical BCG therapy [17‐20]. The side-effects of BCG are common and can be severe; hence, new bladder preservation approaches for HGNMIBC should aim to provide better quality of life (QoL). However, clinical data on the different kinds of intravesical instillation, chemotherapy, radiotherapy, device assisted therapy, and electromotive drug administration is limited [21]. In addition, BP may cause inadequate treatment due to the delay of radical cystectomy (RC). About one third of patients treated with intravesical BCG still undergo RC during the treatment procedure [22]. Delay in RC may increase the risk of lymph node metastases and even bladder cancer-specific mortality [23, 24]. Thus, timely RC has to be performed during HGNMIBC treatment.

As HGNMIBC has a risk of recurrence and progression, RC has become a popular therapeutic strategy [25, 26]. There were multiple studies demonstrating that RC is the preferred treatment option due to better survival rates [27‐29]. However, there is a consensus that RC and intravesical BCG shares the same curative effect and several studies have demonstrated that RC is therapeutically excessive [30]. The RC procedure includes surgically removing the whole bladder along with adjacent organs and reconstructing the urinary drainage, which could cause severe complications and even death. Several studies have reported that many patients suffered from different short-term complications which require more than 6 months for the QoL to normalize to preoperative levels [31]. Because survival and QoL outcomes remain uncertain, the impact of RC still needs to be explored further.

Major associations share similar treatment opinions for HGNMIBC. American Urological Association (AUA) guidelines [32] and European Association of Urology (EAU) guidelines [33] strongly recommends 6 weeks of BCG intravesical instillation for high-risk NMIBC and a maintenance schedule of up to 3 years. The National Comprehensive Cancer Network (NCCN) guidelines [34] also regard BP as the first line treatment option for the management of high-grade tumors. Specifically, observation and intravesical instillation (BCG and chemotherapeutics) should be selected for HG cTa patients whereas BCG is the only recommended BP modality for HG cT1 patients. As for RC, EAU, and NCCN, their guidelines recommend that immediate or early (within 3 months of diagnosis) RC should be considered as an option worth discussing with a low grade of evidence. More specifically, AUA guidelines for grade C suggest after a single course of BCG, RC should be offered for high-grade T1 patients who are fit for surgery and a second course of induction should be considered for persistent or recurrent high-grade Ta or CIS patients. However, NMIBC could be a mixture of CIS, Ta, and T1 instead of containing only one single grade of tumor. Although consensuses exists among the three guidelines, questions regarding BP selection, necessity and timing of RC, and the selection of optimal treatment strategies between BP and RC still remain controversial [35, 36].

To help determine the optimal treatment modality, we performed a meta-analysis of BP versus RC for patients with HGNMIBC. The survival outcomes included overall survival (OS), cancer-specific survival (CSS), and progression-free survival (PFS).

Intravesical BCG was first used as a treatment for superficial bladder cancer in 1976 [47]. Since then, this treatment has gradually become the gold standard and stood the test of time for decades. Nowadays, several new bladder preservation approaches have emerged to replace BCG by providing comparable therapeutic efficacy and better quality of life. However, with reference to HGNMIBC regarding high risk of recurrence, progression, and disease-specific death, radical cystectomy still cannot be ruled out during the treatment selection procedure.

In order to provide convincing therapeutic evidence, we performed this meta-analysis to compare BP and RC in HGNMIBC patients. HR was used to compare treatment modality risk through time. The 2-year, 5-year, 10-year, and 15-year OR were used to assess the short-term to long-term survival outcome.

Firstly, all OR and HR of OS reflected significant difference in favor of BP as recommend in AUA, EAU, and NCCN guidelines. As for the subgroup analysis, approximately all groups showed post-BP patients having better survival compared to post-RC patients except for RC patients younger than 65 years old in 5-year and 10-year OS where overall survival advantage versus BP was not statistically significant. However, only several subgroup analyses showed significant differences supporting BP including the G3 group (5-year OS) and > 65-year-old group (5-year and 10-year OS). Hence, we concluded that BP was a more effective treatment modality for HGNMIBC especially for older patients and patients with G3 state tumors. The recommendation has been made that BC patients with failure of local therapy or progression should receive BP following delayed RC rather than immediate RC [48]. The superiority of mixed BP modality compared to RC for 5-year OS could indicate the inferiority of BCG, yet the strength of the evidence is very low. Regarding the original tumor state, the difference in mid-term overall survival may be affected only in T1 patients in the G3 group.

Unlike the OS analysis, the results of CSS failed to confirm any significant difference except for the 15-year CSS, which suggested RC was a better management option for long-term results. The different results between OS and CSS could be explained due to RC causing non-BC mortality. We observed that OR of CSS gradually increases over time. This result may indicate that the complication of surgery has a great impact and decreases over time. A previous long-term study reported 28% of early complications for RC and caused 2.5% of peri-operative deaths [35]. Including from what we observed from the OS results, RC may be a better treatment option for younger patients with better life expectancy, while BP is a better option for elderly patients [49]. The subgroup analysis of radical cystectomy timing showed that RC surgery performed after more than 3 months from diagnosis was a better treatment opinion compared to BP in the subgroup analysis of HR and OR (2-year and 10-year CSS) in the same single study. Therefore, these results may further prove the advantage of BP following a delayed RC. BCG was considered a better option compared to mix of BP modality, i.e., compared to RC separately in the 10-year CSS, which was opposite to the result of the 5-year OS; however, both had a low degree of evidence strength. But this paradoxical result could be explained by the non-cancer-specific death in RC as well. Although a meta-analysis reported that no statistical significance was found in survival outcome between BCG and MMC [50], recent randomized controlled studies have confirmed that BCG has a better recurrence prevention compared to epirubicin [51], epirubicin+interferon [52], or MMC [53]. Hence, BP modality selection is still complex and unclear. Hence, we may need to weigh the advantages and disadvantages of different BP before selection. Regarding original tumor state, RC showed a statistically significant better result for the 10-year CSS in the HG group. This could be explained by CIS, as a confirmed risk factor that decreases survival rates in high-grade tumors [54]. Hence, a more aggressive treatment modality like RC may be more suitable but has to be confirmed in more studies.

Intravesical BCG and RC are considered progression-reducing modalities in separate studies for BC [55, 56]. When comparing the two treatment modalities, the HR and OS results in PFS showed no significant difference with all subgroup analysis having limited quantities due to lack of data. Hence, both BP and RC may have the same effect on preventing disease progression.

In addition to survival outcome, quality of life was mostly affected by the side-effects and complications of treatment and is a crucial determinant for deciding a management modality. Body image and emotional and psychosocial stress are well-known psychosocial side-effects of RC, and more than 80% of patients receiving RC complained of sexual dysfunction [57, 58]. More importantly, severe complications such as fistulas often require surgical repair and sometimes could result in death [59, 60]. As for BP, the side-effect of BCG are usually mild; however, less than 5% of patients can present with severe systemic complications and may even be life-threatening [61]. When choosing the optimal management for HGNMIBC, both survival outcome and QoL outcome need to be weighted according to patients’ characteristics and life expectancy. In addition, the necessity of close follow-up is unarguable and recommended by AUA, EUA, and NCCN guidelines. Even for RC patients, strict long-term monitoring and continuous follow-up are indispensable for monitoring high-risk recurrence and progression [13]. Early detection following early treatment will lead to better survival outcome.

Several significant heterogeneities were observed which could be explained by the differences among studies such as study size, timing of radical cystectomy, bladder preservation modality, and follow-up time. Generally, a larger sample size or longer follow-up time represented a more stable outcome. Among the included studies, heterogeneity could be a result of radical cystectomy performed within or more than 3 months after diagnosis and bladder preservation modality, which varied from study to study. Characteristics of tumor (unifocal or multifocal, size, and grade) and patients (race, age, gender, and risk classification) could also contribute to the heterogeneity results. Multifocal, larger size, and higher grade result in a greater likelihood of a malignant behavior. Differences in race, age, gender, and risk classification may also cause a difference in disease feature and therapeutic responsiveness. Older patients and high-risk tumors usually have the worse prognosis. In addition, the difference in skill level of physician and pathologist will strongly affect the diagnosis and treatment and needs to be considered as an influencing factor.

There were several limitations to our analysis. First, no randomized control trials were found during our searching process probably because of ethical reasons; hence, only included retrospective cohort studies were used for our analysis. Second, differences among eligible studies were inevitable. We selected studies conducted worldwide; hence, the race of the patients were variable. Multiple therapeutic agents and management schedules were used in the BP group. In addition, the observation was included which may cause the BP effect to vary from study to study. Additionally, the use of either the 1973 WHO grading system or 2004 revision may have affected the results of our meta-analysis. Third, as outlined by the NOS scale, some of the included studies failed to control additional factors for comparability, had a short-term follow-up, or did not describe the follow-up schedule and hence contributed to the bias of the studies. Fourth, due to missing or lack of data, some analyses were performed on limited studies, and several subgroup analysis were unable to be performed. Fifth, disease recurrence is a very important outcome but no study provided such data for analysis.

From our meta-analysis, we concluded that although having the same effect of preventing cancer-specific death and progression from HGNMIBC, bladder preservation approach is a superior modality compared to radical cystectomy. It provides better overall survival outcome, especially for the elderly and patients with T1G3 tumor. However, choosing an optimal BP strategy is still unclear and indefinite. For patients with expected longer life expectancy, RC could be a better option (more specifically, RC performed more than 3 months from diagnosis). However, this conclusion is from very limited data, which needs to be verified using future studies. During the decision-making process of selecting a treatment modality, quality of life should be considered equal to survival outcome. Last but not the least, a close follow-up is essential for any treatment modality.

Heterogeneity and limitations were inevitable and decreased the reliability of our study. Hence, a high quality, prospective, randomized study comparing the effectiveness and the tolerability of BP versus RC is extremely necessary in the future. There are protocols for randomized controlled feasibility studies designed to compare intravesical BCG and RC for high-risk non-muscle-invasive bladder cancer [62]. With the development of multiple BP modalities and reducing surgery, associated death rate will help develop an excellent curative effect and quality of life.