Complex renal cysts (Bosniak ≥IIF): interobserver agreement, progression and malignancy rates

A regional database stores the health records of patients with complex renal cysts diagnosed in a defined geographical area with a population of more than 416,090 people, based on mid-year 2017 population estimates [15]. The stable population with less than 1% migration rate within the area is registered with the health board through a unique 10-digit identifier called a Community Health Index (CHI) number. This is a further update to our previous publication, where the process of cohort identification has been described [7]. Briefly, all patients with renal cysts diagnosed between January 2009 and December 2019 were recorded in the regional database and reviewed at multidisciplinary meetings. The discussions were recorded on a proforma (Appendix 1), and patients were followed up with interval imaging as necessary. Following consensus at multidisciplinary meetings, category I and II cysts were discharged and the remaining were followed up. In the case of radiological progression, further discussions in multidisciplinary meetings were arranged and treatment plans recorded. For the purposes of investigating complex cysts, Bosniak I and II cysts, haemorrhagic cysts, inflammatory cysts and small renal masses (SRMs) were excluded from analysis. Figure 1 illustrates the study design.

The data from each patient with renal cysts were documented on a secure electronic database and linked to other electronic databases using a unique 10-digit Community Health Index (CHI) number. The databases linked using the CHI number included the National Picture Archiving and Communication System (PACS), Clinical Portal, Integrated Clinical Environment (ICE), Referral Management System (RMS) and Electronic Discharges from hospital systems.

The deterministic records-linkage methodology between systems provided us with the time to follow-up, time to progression/regression, details of surgical intervention and the data of the resected specimen such as histologic subtype, grade and stage.

There is an established protocol to characterise cysts at our institution as described before [7]. The CT renal characterisation scan is triphasic with precontrast, arterial (40 s post-contrast) and nephrographic phases (100 s post-contrast) following IV administration of 100 ml of iodine-based contrast agent (Omnipaque) and acquired as a volume using a 64-slice thickness CT scanner (Fig. 2). The 100-s delay helps to visualise the kidneys in their nephrographic phase, the optimal phase to characterise renal cysts. The nephrographic phase was also obtained using MRI and gadolinium contrast. T2 HASTE (6-mm slice thickness, 1.8-mm gap), T1 VIBE pre- and post-contrast (3.5-mm slice thickness, 0.7-mm gap), STIR (8-mm slice thickness, 2.4-mm gap) and diffusion-weighted (4-mm slice thickness, 1.2-mm gap; b values 50, 400, 1000 s/mm²) images were obtained. The change in cyst enhancement was measured by calculating the difference in pre- and post-contrast enhancement.

Patients with Bosniak IIF cysts were followed up in accordance with the local protocol with serial imaging at 6, 12, 24, 36 and 48 months, preferably using the same imaging modality. In cases of high likelihood of contrast nephropathy, MRI was used for surveillance. The follow-up period was adjusted subject to a patient’s co-morbidities and suitability for surgical intervention. If the follow-up imaging demonstrated an increased complexity of the cyst, the case was referred to the multidisciplinary meeting and the imaging was reviewed by the uroradiologist. Cyst progression was defined by increased cyst complexity resulting in an upgrade in Bosniak category. Radiological findings attributable to cyst progression included progressing nodule enhancement, septal thickening and wall thickening [2].

Patients with Bosniak III and IV cysts were discussed in clinic and surgical intervention was recommended. Surgical intervention included partial or total nephrectomy, radiofrequency ablation or cryoablation, subject to cyst characteristics and patient-specific factors. Poor surgical candidates or those who opted for conservative management were followed up with serial imaging or were discharged, subject to patient preference.

The interobserver agreement was evaluated by comparing the Bosniak category reported on the initial scan report at the time of detection, with the category reported at the multidisciplinary meeting by the specialist uroradiologist. All cysts discussed at the multidisciplinary meeting (Bosniak ≥II) and reported using the Bosniak classification were included. The linear and quadratic-weighted kappa statistics were calculated to evaluate the rate of agreement.

There were 257 cysts categorised on the initial radiology report and followed up at a multidisciplinary meeting (Table 1). The two readers agreed on the classification in 70% of cases (179/257). Overall, there was a good degree of agreement between the two independent radiologists for all Bosniak cysts ≥ II, as demonstrated by a linear and quadratic-weighted kappa statistic of 0.65 (95% confidence interval, 0.58–0.73) and 0.73 (95% confidence interval, 0.54–0.91), respectively.

The rate of surgical intervention for IIF, III and IV cysts was 3.1% (5/161), 36.7% (29/79) and 59.7% (46/77), respectively. Of the resected cysts, 6.3% (5 cysts) were progressed IIF cysts, 36.3% (29 cysts) were Bosniak III cysts and 57.5% (46 cysts) were Bosniak IV cysts. In 53% of cases, a partial nephrectomy was performed; in the remaining 47%, a total nephrectomy was performed.

The overall rate of malignancy in surgically resected cysts was 81.3% (65/80) (Table 4). Of the malignant tumours, histopathology confirmed that 73.8% were clear cell, 13.8% were papillary, 6.2% were multilocular cystic carcinomas and the remaining were either chromophobe or collecting duct carcinomas. The malignancy rate in cysts that were resected after a delay versus those resected immediately was 76.9% (10/13) and 82.1% (55/67), respectively, and this was not statistically significant. Of all malignant tumours, 73.8% were of low ISUP grade, 93.7% were confined to the kidney (stage pT2b or less) and 81.0% were less than 7 cm (stage pT1a or pT1b). The recurrence rate was 1.5% (1/65) after a median follow-up time of 50.2 months (range, 2.8–93.1 months). The grade and stage of the tumours were not statistically different between those who had immediate or delayed surgery.

Overall, there is a good degree of agreement between radiologists using the Bosniak classification system; however, Bosniak II cysts are frequently over-graded. Both the rate of progression and regression of Bosniak IIF cysts are low (4.6% and 3.3%, respectively), and malignant cysts progress during the early surveillance period (within 16 months). Surgically resected Bosniak III and IV cysts have a high rate of malignancy, but the rate of benign cyst resection is still high (18.8%). Malignant cysts on histopathology showed low-grade and early-stage renal cancers.

The consensus from the literature is that a surveillance period is safe and facilitates identification of Bosniak IIF lesions that require surgical intervention [9]. A definitive surveillance period for Bosniak IIF cysts has not been settled upon within the literature, but a period of up to 4 years has been suggested [1, 6, 9]. Our data suggests that malignant progression and regression in most cases occur early in the surveillance period and a shorter surveillance period would be acceptable. Of course, the possibility of later malignant progression remains, albeit very small and was not observed in this population.

There is no universally accepted interval period between the surveillance scans for Bosniak IIF cysts [16]. In our study, the Bosniak IIF cysts were followed up in accordance with the local protocol with serial imaging at 6, 12, 24, 36 and 48 months, subject to patient factors. A total of 287 follow-up scans were performed on these cysts, and seven instances of progression were reported. Although the malignancy rate of progressed IIF cysts is high, all three of the malignant Bosniak IIF cysts were early stage and low grade; thus, a greater interval between scans may be appropriate [17]. One study suggests that a longer interval period permits cysts sufficient time to develop radiological features and reduces radiation exposure [18]. The rate of progression of Bosniak IIF cysts within the literature ranges from 4.6 to 15.6% [5‐7]. In the present study, the progression rate was low (4.6%) and supports a shorter follow-up period and a greater interval time between surveillance scans.

The unnecessary resection of benign Bosniak III cysts is a recognised limitation of the Bosniak classification. Prior to the introduction of the IIF category, the malignancy rate of Bosniak III cysts had been reported as low as 31% and 45% [3, 11]. The addition of the intermediate IIF category has improved the clinical significance of the Bosniak classification, as evidenced by increased malignancy rates of Bosniak III cysts and decreased resection rates. Studies have since reported malignancy rates between 60 and 81.8% [8, 19]. Graumann et al conducted a meta-analysis of 15 retrospective studies and found that the malignancy rate was 65.4% [20]. In the present study, we report a malignancy rate of 79.3% and six resected Bosniak III cysts (20.7%) were benign. The revised Bosniak classification published by Silverman (2019) et al aimed at reducing the frequency of benign cyst resections by incorporating explicit definitions of terms and specific inclusion criteria. It defines the number of septae, thickness of wall and septae and nodularity for each category and sets the criteria to increase the specificity of the classification system [21]. These features were not tested in the present study, and whether the revised Bosniak classification of 2019 improves the specificity of categorisation in predicting malignancy remains to be seen. We envisage that adoption of more qualitative and quantitative criteria with machine learning algorithms would certainly reduce the frequency of benign cyst resections, improve our ability to follow up these lesions on active surveillance and increase the malignancy rate of resected cystic renal masses. Future research is required to externally validate the 2019 classification.

The ISUP grading system is employed to predict the biological aggressiveness of the cancer, and an ISUP grade of 1 or 2 predicts a relatively indolent cancer. Of the resected Bosniak III and IV cysts, 89.5% and 69.2% were of low ISUP grade, respectively. These findings correlate with a low grade of renal cell carcinomas in the literature irrespective of Bosniak category [13, 14]. Given the relatively indolent nature of the malignant cysts, survival analysis studies with long-term follow-up would be useful to identify patient sub-groups who would benefit from surgical intervention or surveillance management, particularly in the Bosniak III and IV groups. Surgical decision-making in the management of complex renal cystic disease should also consider patient-related factors such as age and number of co-morbidities in addition to the risk of malignancy based on classifications of structural abnormalities on imaging. Any radiological classification can only predict risk of malignancy and not biological behaviour of renal cancers.

In our study, there was no correlation between delayed surgery and higher grade. All cysts that underwent delayed surgery were of low grade, and the period of observation did not compromise outcome. A period of observation may help in selecting progressing cysts, increase the malignancy rate of resected Bosniak III cysts and reduce the proportion of benign cyst resections. Further studies are warranted to investigate if a period of observation would reduce the rate of benign surgery and not compromise cancer outcomes.