Results of the IROCA international clinical audit in prostate cancer radiotherapy at six comprehensive cancer centres

To assess adherence to standard clinical practice for the diagnosis and treatment of patients undergoing prostate cancer (PCa) radiotherapy in four European countries using clinical audits as part of the international IROCA project. Multi-institutional, retrospective cohort study of 240 randomly-selected patients treated for PCa (n = 40/centre) in the year 2015 at six European hospitals. Clinical indicators applicable to general and PCa-specific radiotherapy processes were evaluated. All data were obtained directly from medical records. The audits were performed in the year 2017. Adherence to clinical protocols and practices was satisfactory, but with substantial inter-centre variability in numerous variables, as follows: staging MRI (range 27.5–87.5% of cases); presentation to multidisciplinary tumour board (2.5–100%); time elapsed between initial visit to the radiation oncology department and treatment initiation (42–102.5 days); number of treatment interruptions ≥ 1 day (7.5–97.5%). The most common deviation from standard clinical practice was inconsistent data registration, mainly failure to report data related to diagnosis, treatment, and/or adverse events. This clinical audit detected substantial inter-centre variability in adherence to standard clinical practice, most notably inconsistent record keeping. These findings confirm the value of performing clinical audits to detect deviations from standard clinical practices and procedures.


Selection of quality indicators and clinical parameters.
First, the working group reviewed the relevant literature 3,11,12 , including the main clinical guidelines (NCCN and EAU) for staging and treatment, as well as institutional and national guidelines. Based on this review and on our previous experience in performing clinical audits 2,10 , we selected a set of clinical indicators most relevant to PCa (supplementary Table S1). Three key phases of care were assessed: (1) (2) curative-intent treatment (EBRT and/ or BT). Exclusion criteria were: (1) absence of radical radiotherapy treatment, (2) presence of recurrent disease at diagnosis, and/or (3) presence of other cancers except non-melanoma skin cancer.
The sample size was estimated by assuming a reference proportion of 50% for any variable, with a minimum difference between two hospitals defined as 25%, with an alpha risk of 0.05 and beta of 0.10. The resulting sample size per hospital was 40 cases. Patient selection. All patients who met the inclusion criteria were assigned an identification number and then randomly selected for participation. After inclusion in the study, the selected patients were then stratified by risk groups (NCCN criteria), as follows: low-risk (≤ stage T2a and Gleason score [GS] ≤ 6 and PSA < 10 ng/ mL); intermediate-risk (T2b-T2c or GS 7 or PSA 10-20 ng/mL), and high-risk (≥ T3a or GS ≥ 8 or PSA > 20 ng/ mL or N+). Patients with multiple adverse intermediate risk factors were not included in the high-risk group if they did not present any high risk criteria. Statistical analysis. The Chi square test for categorical variables was used to compare results among hospitals. Fisher's exact test was used when frequencies were < 5. The SPSS-IBM statistical software program, v.21 (IBM, Armonk, NY; USA) and R software version 3.6.3 were used to perform the statistical analyses. Table 1 shows the sample characteristics. Most patients (n = 138; 57.5%) were between 66 and 75 years of age. The biopsy report was included in the medical record in 97.9% of cases. Most patients (60.4%) underwent staging MRI of the prostate gland and pelvic lymph nodes, although inter-centre variability was wide (range 27.5-87.5%). CT scans were performed in 40.4% of the patients, mainly high risk ones, and some of these underwent PET scan (data not available). Bone scans were performed in 58.3% of the patients (inter-centre range 47.5-67.5%). Overall, 51.7% (n = 124) of patients were presented to the MTB, with significant (p < 0.001) inter-centre variation (range 2.5-100%). Two institutions failed to indicate on the medical records whether the patients were presented to the MTB.

Results
Risk group classification (NCCN criteria) was as follows: high risk (n = 120; 50%), intermediate risk (31.7%), and low risk (18.3%) (supplementary Figure S1).  Table S2). Overall, most cases (62.5%) were presented at the departmental clinical session prior to treatment initiation; however, inter-institutional variability was large. Two centres (ICO-B and ICO-G) only presented selected cases at the departmental session.
Overall, median time between the first visit to the radiation oncology department and treatment initiation (EBRT or BT) was 78 days (range 42-102.5). Median time between CT simulation and radiotherapy was 18 days (range . Most patients (87.1%, n = 209) received EBRT+/− a BT boost (n = 22). The remaining 31 patients (12.9%) received BT alone (supplementary Figure S2).
A wide range of doses were administered to the prostate in patients undergoing EBRT (supplementary Figure S3), mainly due to the use of different treatment approaches and combinations, which vary according to the risk category (low, intermediate, high).
The dose per fraction (Gy) in patients treated with EBRT is shown in supplementary Figure S4. Dose constraints were met in all cases unless otherwise indicated. Three of the six participating centres offered BT (supplementary Table S3). Fifty-three patients received BT, either HDR or LDR; in 17% of these patients, the dose rate was not specified in the medical record. Thirty-one patients received BT as monotherapy (145 Gy): 25 received  2 and 3 show the treatment distribution according to risk group, revealing a wide range of different treatment regimens, consistent with routine variability in the radiotherapeutic treatment of PCa. As Fig. 1 shows, a range of different treatment regimens were used in the low-risk patients (n = 44). However, EBRT alone   www.nature.com/scientificreports/ (n = 23; 52.3%) and BT alone (n = 17; 38.6%) accounted for 90.9% of all low-risk treatments. Only 4 patients received hormone therapy (HT). Seventy-six patients (31.7%) had intermediate-risk PCa (Fig. 2). The most common treatment modalities in this risk group were long-term androgen-deprivation therapy (ADT) + EBRT (n = 29, 38.2%), followed by EBRT alone (n = 22, 28.9%), short-term ADT + EBRT (n = 13, 17.1%), and BT alone (n = 9, 11.8%). The remaining treatment options were used in only a limited number of cases. Intermediate-risk PCa, which comprises a heterogenous group of patients, is now divided into favourable and unfavourable intermediate-risk groups 13,14 . However, when these patients were treated, intermediate-risk patients were grouped into a single category (2016 NCCN criteria), and thus our intermediate group contains patients with both favourable and unfavourable risk characteristics, some of whom could be considered either unfavourable intermediate or high risk, depending on the clinical guidelines (i.e., NCCN or EUA). Figure 3 shows the treatment distribution for high-risk patients. Exactly 50% of patients in this study (120/240) had high-risk PCa. The most common treatment regimens in the high-risk group were long-term ADT + EBRT (n = 86, 71.7%), long-term ADT + EBRT + BT boost (n = 10, 8.3%), and short-term ADT + EBRT + BT boost (n = 10, 8.3%). Only 2 patients (1.7%) received EBRT alone, probably due to the presence of comorbidities (e.g., cardiovascular disease) that contraindicated ADT (data not registered).
Treatment-related indicators are shown in supplementary Table S4. Treatment interruptions ≥ 1 day occurred in 67.1% of the patients treated with EBRT, ranging from 7.5 to 97.5% of cases, depending on the treating centre (p < 0.001).
Compensation rates for treatment interruptions were highly variable (range 0-87.5%), with some centres compensating for most or all interruptions while other centres either did not compensate for interruptions or data were not reported. Overall, 27.3% (range 5-72.4%) of patients completed treatment in the prescribed time; however, due to the relatively indolent nature of most cases of PCa, treatment interruptions of ≤ 4 days are considered reasonable and unlikely to affect the outcomes 15 . The percentage of patients completing treatment in the prescribed time ± 4 days was 67% (range 43.3-100%).
For image guidance, we assessed whether image-guided radiotherapy (IGRT) was performed and the technique (cone-beam CT [CBCT], fiducial markers, and KV/MV). All methods except KV/MV were considered acceptable (22% of patients).
Three of the six centres ICO-H, IPO and WCO) performed BT; in two of these centres (ICO-H and WCO), the BT boost was performed in 54% of high-risk patients. A BT boost was administered in 17.5% (n = 21) of high-risk patients overall.
Adverse effects. AEs in this sample are summarized in supplementary Table S5. All participating centres registered AEs, which were observed in 65.4% of patients. The remaining patients (34.6%) either experienced no AEs or these were not registered.

Follow-up.
Most patients (96.7%) were alive at last follow-up. Regular follow-up (defined as ≥ 2 visits per year) was performed in nearly all patients (97.9%).

Discussion
The present international, multi-institutional clinical audit was performed to compare adherence to standard clinical practice for prostate radiotherapy across six European cancer centres. Based on the predefined audit criteria, the participating centres generally adhered to standard clinical practice for diagnosis, treatment, and follow-up of PCa. Nonetheless, there was substantial inter-centre variability in clinical practice and several minor process-related deficiencies were detected. The key findings of the clinical audit are presented in Table 2 and discussed below.
Slightly more than half of the cases were presented to the MTB, with substantial inter-centre variation (2.5-100%). This finding was not unexpected given the lack of a generally-accepted protocol for presenting patients to a MTB, and these heterogenous findings appear to reflect variability in clinical practice across Europe. Due to missing data, we were unable to determine whether 56 cases (23.3%) had been presented to the MTB. Although this high number of missing cases suggests a failure to register key data, a large proportion of these "missing" cases are likely low-risk patients who underwent BT monotherapy, which are not routinely presented to the MTB in some centres. The value of presenting all cases to a MTB prior to treatment is supported by published reports showing that the diagnosis and/or treatment plan can change in up to 50% of cases after presentation to the MTB 16 . In a survey of patients with non-metastatic prostate cancer conducted by Patrikidou et al. 17 , multidisciplinary consultation improved patient satisfaction and strongly influenced the final treatment decision. However, the structure of the MTB and the expertise of the participants play a vital role in the value of a given MTB 18 . Considering the time and expense involved in routinely presenting patients to a MTB, some authors recommend "mini" tumour boards comprised of only a few selected experts when there is a lack of available expertise and/or specialists 19,20 .
Four of the six hospitals (66.7%) in this study presented all or nearly all patients at the department clinical session prior to treatment initiation. However, two of the centres reported presenting only complex or unique cases. During the post-audit meeting to discuss our findings, representatives from the participating institutions debated the value of routinely presenting all patients at the departmental session. The centres that do not www.nature.com/scientificreports/ regularly do so argued that heavy caseloads and time constraints make this impractical and-moreover-may not provide much value due to the scant time allotted to each individual case; in addition, if a large number of routine cases are presented, clinicians are less likely to pay attention. By contrast, other researchers argued that all cases should be presented as a quality control measure to detect errors and to help train residents. One centre takes an intermediate approach, presenting special cases to a "mini-board" comprised of specialists in that specific tumour site, an approach that offers several potential advantages: it avoids "information overload" at the departmental clinical sessions and ensures that physicians who evaluate the case have specific expertise in that tumour site. One key measure of quality is the inclusion of all relevant clinical data in medical record 21 . In patients with cancer, the biopsy report provides essential information. Unsurprisingly, we found that the biopsy report was included in the medical records of 98% of cases, indicating good adherence to quality among the participating institutions. Nevertheless, the few missing cases point to a minor but important deviation from good practice.
Most clinical guidelines 22 recommend staging MRI for patients with PCa, although its value in low-risk patients (except in active surveillance) has been questioned [22][23][24] . Most patients (60.4%) in this study underwent staging MRI, but with wide inter-centre variability (27.5-87.5%), possibly due to difference in clinical practices. For example, the centre with the lowest proportion of patients who underwent staging MRI used an alternative technique (ultrasound) to stage patients with low-risk PCa and MRI/CT fusion in intermediate and high-risk patients. CT can be considered acceptable to detect metastatic disease, but not for prostate gland evaluation. Finally, although staging MRI has now become standard, during the study period (2015) many centres were still transitioning from CT.
Waiting times are among the most important quality variables, as excessive delays between diagnosis and treatment initiation can negatively impact treatment outcomes by increasing the risk of tumour growth and metastasis 25 . Although longer waits are unlikely to impact clinical outcomes in patients with low-risk PCa, this is not true in intermediate or high-risk cases, with the available evidence showing that excessive waiting times can negatively affect oncological outcomes 26 . For this reason, it is crucial to minimize waiting times whenever possible. We observed wide variability in the time elapsed from the first visit to the radiation oncology department to the start of radiotherapy (range 42-102.5 days). Several factors may account for these inter-institutional differences, including differences in patient management practices: in some centres, staging is performed by the radiation oncologist (which could delay treatment initiation) whereas in other centres the patient is referred to the radiation oncologist after staging has been performed. In some cases, treatment may be intentionally delayed when there is a clinical advantage in doing so (e.g., in patients with highly symptomatic disease, or in those who recently started HT). The median time interval between CT simulation and radiotherapy was 18 days (range 12-33 days). One centre presented a wider range (16-120 days), suggesting poor compliance with standard clinical practice, which recommends that the CT be repeated after ≥ 30 days if the patient has not-for any reason-initiated radiotherapy 22 . www.nature.com/scientificreports/ Studies conducted to assess the causes of long waiting times suggest that lack of timely treatment may be related to inefficient healthcare processes, limited medical resources, and absence of policy support 27 . Some authors have suggested that these should be considered an indicator of access to healthcare 28 . Importantly, longer waiting times are often-although not always-associated with worse outcomes 29,30 . Wyatt et al. found that a delay of 1-2 months does not significantly impact tumour control rates in patients with PCa 31 . By contrast, Nakayama et al. found that delays > 6 months increased the risk of biochemical progression in patients with localized PCa 32 . Mackillop et al. 29 concluded that delayed initiation of radiotherapy may be associated with a clinically-significant deterioration in local control, leading those authors to emphasize the importance of minimizing these times. In this context, our findings reveal that most, but not all, of the participating centres complied with clinical protocols for this indicator. Clearly, the centres that failed to comply with these recommendations will need to take measures to ensure that waiting times are within acceptable limits.
Our results show wide variability in treatments, even among patients in the same risk category. Most of the low-risk patients underwent single-modality therapy-either EBRT alone (52.3%) or BT alone (38.6%)-data that are in line with current clinical guidelines 33 . As might be expected, the widest variability in treatment selection was observed in the intermediate-risk group (Fig. 2), even though most (67%) of those patients were treated either with EBRT alone (28.9%) or EBRT + long-term ADT (38.2%). Nevertheless, this variability among intermediate-risk patients should be expected due to differences within this risk group, as some may be closer to low-risk (e.g., those who received EBRT alone) while others present features closer to high-risk patients. Intermediate-risk PCa is highly heterogeneous, displaying a wide range of clinical behaviour and aggressiveness, and biochemical and clinical recurrence rates can range from as low as 2% to as high as 70% 34,35 , which is why intermediate-risk patients are now generally subclassified into favourable vs. unfavourable risk groups to more accurately reflect the real risk of recurrence 33,[36][37][38] .
Treatment interruptions are an important indicator of quality that can have a marked effect on outcomes 39 . The usual causes of interruptions include machine malfunction or maintenance, toxicity, holidays, or unplanned quality control checks, which is why it is essential to have in place protocols to determine how to compensate for the interruption. In most cases, definitive EBRT with conventional fractionation schemes for localized PCa requires 8 weeks of treatment, which means that treatment interruptions are common in these patients, even though brief interruptions do not appear to have a major impact on outcomes. Dong et al. 15 found that even long treatment breaks (≥ 4 fraction) did not significantly affect survival outcomes in patients receiving high dose (> 74 Gy) radiation without ADT, provided that all missing fractions were delivered by the end of treatment. As those authors observed, those findings reflect the relatively indolent course of PCa compared to other cancers. Nonetheless, quality clinical practice dictates that all efforts should be made to avoid interrupting radical radiotherapy treatments; when interruptions are unavoidable, compensation is essential. According to the Royal College of Radiologists 40 , any interruption to radiotherapy schedules may affect the outcome and thus interruptions should be "as short as reasonably achievable". Although prolongation up to 5 days may not be detrimental, no safe minimum has been established, which is why those guidelines recommend limiting treatment prolongation to no more than 2 days over the original prescription 2,40 .
This clinical audit revealed important differences among centres in treatment interruptions, raising the possibility that some institutions may need to re-assess quality control for this particular aspect. Treatment interruptions of 1 day or more were common, occurring in two-thirds of patients treated with EBRT, with significant variability among centres (7.5-97.5%). Similar inter-centre heterogeneity was observed with regard to treatment compensation rates. While some centres compensated for all (or nearly all) interruptions, others either did not do so or-importantly-data were not available. Compensation rates ranged from 0 to 87.5% (overall, 9.9%). Surprisingly, only slightly more than one-quarter of patients who underwent EBRT completed treatment in the prescribed time, with high variability among centres (5-72.4%). However, if we add 4 days to the expected overall treatment time (OTT)-a common practice in routine care-the percentage of patients completing treatment in the prescribed time rises to 67% (range 43.3-100%). Ideally, 100% of patients should complete treatment in the prescribed time and the 67% rate in this study points to a clear target for improvement. The centres with the highest percentage of treatment interruptions were in Spain and Italy, with the lowest in Poland and Portugal. While the reasons for these differences are not entirely clear, a recent single-institution study in Spain 41 found that public holidays accounted for 45% of unscheduled interruptions, and these interruptions were associated with a lower probability of local disease control and/or an increase in the likelihood of biochemical failure in patients with PCa. Based on those findings, the authors concluded that outcomes of fractionated radiotherapy could be improved by working on public holidays. That same group also reported their experience in managing interruptions 42 , recommending two concrete measures to reduce OTT: working on public holidays and performing maintenance tasks on Saturdays. Those authors calculated that implementation of those two measures reduced the likelihood of biochemical failure by a mean of 2% (and by ≥ 4% in nearly 20% of patients). Other authors have also provided advice on how to manage unscheduled treatment interruptions in patients with PCa 43 .
Clinical recommendations call for long-term ADT (≥ 2 years) in high-risk patients. In our cohort, 96 of these 120 high-risk patients (80%) received long-term ADT, although inter-institutional variability was high, suggesting less than optimal adherence to clinical guidelines; however, because ADT is contraindicated in patients with certain conditions (e.g., severe cardiovascular disease), it seems likely that at least some of the patients who did not receive long-term ADT may have presented some type of contraindication.
We found that all centres registered AEs on the medical record, with 65.4% of patients experiencing AEs and the remaining cases (34.6%) either not presenting AEs or these events were not registered. Clinical guidelines stipulate that the presence or absence of side effects should be reported 22,23,33 . In this regard, it appears that AEs may have been underreported at some centres and/or differently graded. For example, one institution failed to indicate in 23 cases whether or not cystitis-urethritis was present. In addition, in 102 cases (42.5%), the participating institutions failed to report the present or absence of erectile dysfunction. This finding shows that only www.nature.com/scientificreports/ some of the acute AEs were registered on the medical record. The failure to register all relevant AEs (or their absence) highlights an area that should be targeted for improvement. Nearly all patients (n = 235, 98%) received regular follow up (≥ 2 visits per year) but due to short follow-up and wide variability in risk groups, it was not possible to compare the outcomes among the participating centres.
Although we have sought to put the findings of the present audit in the context of other studies, it is difficult to compare results due to methodological differences, particularly due heterogeneity in the parameters evaluated. Comparison is further hindered by the scant data available and because the parameters must, of necessity, vary according to tumour type, as evidenced in the previous IROCA studies for prostate and rectal cancer 9,44 . In this regard, we still lack a set of internationally recognized, standardized indicators to permit international comparisons among radiotherapy centres 3 , despite efforts made by several different groups [3][4][5]11,45 .
Most clinical audits conducted to date in radiotherapy have been single-centre audits, many of which have been performed using the IAEA audit methodology known as Quality Assurance Team for Radiation Oncology (QUATRO). Izewska et al. 46 explored the factors that impacted quality of care among QUATRO-audited centres (n = 34) in Europe. The main findings relevant to barriers to quality care were insufficient staffing, education/ training, equipment, and lack of quality management. Few multi-institutional audits have been carried out to date, with the notable exception of those performed by our group and by Dutch and Belgian groups 6,7,47 . Aside from the IROCA project, one of the largest multi-centre audits was a national systematic clinical audit of all radiotherapy departments in Belgium (n = 25) from 2011 to 2015 using the IAEA QUATRO methodology. In that study, the authors analysed the clinical audit reports for the individual departments to identify the most common areas for improvement, which included the need for process improvement and protocol development (a recommendation found in 34% of the reports), deficiencies in human resources and physical infrastructure (27%), issues related to departmental organisation (20%), and no systematised approach to follow-up after radiotherapy delivery. Although those findings echo many of the findings in our audit, the methodological differences make these studies difficult to compare directly.
The main limitation of the present study is the sample size (40 patients/centre), thus limiting the statistical power of the analysis. Although this sample size is insufficient to assess clinical outcomes (e.g., survival), this was not a study aim. Rather, since our aim was to identify differences among the centres in adherence to key processes such as registration of clinical data, the sample size was sufficient to detect numerous aspects amenable to improvement and harmonization ( Table 2). Another limitation was the omission of several potentially relevant variables and/or indicators such as some staging tests (e.g., pelvic CT and bone scintigraphy) and pelvic node irradiation, which were intentionally omitted from the analysis for feasibility reasons. Another limitation was the number of intermediate-risk patients (NCCN risk criteria) who were prescribed long-term ADT by the referring hospital, which impeded our assessment of adherence to standard clinical practice. Although we assessed whether IGRT was performed and the type of IGRT, we did not evaluate the frequency of verification, nor did we comprehensively investigate the method used (e.g., online vs. offline). The main strength of this study was the multi-institutional, multinational design with objective external auditors. Another strength is that, in an effort to minimize bias, we considered only objective data obtained from patient medical records.
The purpose of performing regular clinical audits is to continuously improve processes to obtain better outcomes. Research has shown that the mere act of identifying differences can induce participants to seek to harmonise processes to eliminate large deviations from the norms. Indeed, this was confirmed during the post-study discussions among the researchers, as the findings of the study clearly revealed numerous targets for improvement and/or harmonization.

Conclusions
The present clinical audit, performed at six radiation oncology departments in four European countries, reveals generally good adherence to standard clinical practice. The wide variability of treatment approaches observed in this study reflects, in part, the substantial range of treatment options with comparable clinical outcomes in prostate cancer. However, some minor deviations from quality clinical practice were detected, most notably inconsistent data registration. We also observed inter-institutional differences in the presentation of cases to departmental clinical sessions and to the multidisciplinary tumour board, as well as differences in the approach to managing treatment interruptions.
A major benefit of this clinical audit is that it allowed us to identify and jointly discuss difference in clinical practice-some of which are justified-and areas of weakness, which should ultimately encourage the participating institutions to implement better practices, thus leading to greater harmonization among centres in the management of prostate cancer.
The IROCA clinical audit project may provide a useful template for other institutions to follow, potentially promoting a wider use of external clinical audits in radiotherapy. Despite the time and resources needed to conduct a thorough clinical audit, identifying and correcting deviations from standard clinical practice is essential to improving patient outcomes.

Data availability
The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.