- To test the hypothesis that spatial distribution of positive cores at biopsy is a predictor of unfavourable prostate cancer characteristics at radical prostatectomy (RP) in active surveillance (AS) candidates.
John-Hopkins (AS criteria)
Memorial Sloan-Kettering Cancer Center
Prostate Cancer Research International: Active Surveillance
positive biopsy zone
University of California, San Francisco
The widespread use of PSA in clinical practice, as well as the introduction of lower PSA thresholds for biopsy have contributed to a significant stage migration in prostate cancer . This has led to an increase in the incidence of small and low-risk tumours, which are unlikely to impair patient survival [2-4]. This changing landscape of prostate cancer has led to concerns about over diagnosis and over treatment. Under this premise, active surveillance (AS) emerged as an alternative approach in men with a low likelihood of disease progression . This approach is based on close monitoring of patients for possible changes in tumour characteristics over time, which may delay or even evade active treatment. Therefore, such management should significantly reduce patient exposure to treatment-related toxicities without compromising cancer control.
Over the last decade, several criteria have been proposed to identify patients with favourable tumour characteristics, who may benefit the most from AS protocols [6-9]. Virtually, all these criteria are based on prostate biopsy results (namely Gleason score, number of positive cores, and the percentage of core involvement), as well as on PSA value and PSA density (PSAD) at diagnosis. However, none of them accounts for the location of positive cores at biopsy as a predictor of unfavourable tumour characteristics. Intuitively, positives cores that involve more than one prostatic zone at biopsy might be associated with a larger and/or multiple tumour or foci, which may denote less favourable tumour characteristics (Fig. 1). However, there is a paucity of clinical reports testing this plausible relationship. To address this void, we tested the hypothesis that the involvement of more than one prostatic zone and/or the presence of bilateral positive cores at biopsy are associated with less favourable tumour characteristics in patient candidates for AS according to several commonly used criteria.
We evaluated data of 4047 patients with prostate cancer treated with radical prostatectomy (RP) and anatomically defined extended pelvic lymph node dissection, between 2000 and 2012, at a single tertiary referral center. Among these, we identified patients who received prostate biopsy at our institution and fulfilled at least one of the four following AS criteria: John-Hopkins (JH) , Prostate Cancer Research International: Active Surveillance (PRIAS) , Memorial Sloan-Kettering Cancer Center (MSKCC) , and University of California, San Francisco (UCSF) . These selection criteria yielded 524 evaluable patients.
A TRUS end-fire probe was used at a variable frequency of 5–7.5 MHz to guide an 18-G needle for prostate biopsy. In all, 307 patients (58.6%) received an extended biopsy scheme, with a mean (median, range) of 13.9 (14, 10–19) cores. In the remaining 217; patients (41.4%) diagnosed between 2005 and 2008 a saturation biopsy was performed instead, with a mean (median, range) of 23.7 (24, 20–26) cores, as a part of a prospective study protocol [10, 11]. In all cases, the prostate was divided into six zones, as follows: right apex, right margin, right base, left base, left margin, left apex (Fig. 1). One or more biopsy cores were obtained from each zone. In patients that underwent a saturation biopsy, additional four cores were obtained from the transitional zone. A biopsy zone was considered positive (PBxZ), when one or more cores in that specific zone were found to be positive for tumour.
Biopsy cores were immediately put on a sponge tissue in six (extended scheme) or seven (saturation scheme) different sandwich cassettes and sent for pathological examination.
Surgical procedures were performed by different surgeons, using standardised techniques [12, 13]. The pelvic lymph node dissection scheme used was previously described . Whole RP specimens were processed using serial step-sections at 3 mm in accordance with the Stanford protocol, and tumour volume was calculated by visual inspection as previously described [15, 16]. Pathological assessment of all the biopsy and RP specimens was performed by four expert uro-pathologists.
Patient information included age at surgery (years), serum PSA concentration (ng/mL), total prostate volume at TRUS examination (estimated by using the prolate ellipsoid formula: width × length × height × 0.52), PSAD calculated as serum PSA value/total prostate volume, clinical stage (cT1c vs cT2), number of positive biopsy cores, biopsy scheme type (extended vs saturation), pathological stage (pT2 vs pT3), pathological Gleason score (≤6 vs 3+4 vs 4+3 vs ≥8), pathological tumour size, tumour laterality at biopsy (unilateral vs bilateral), and number of PBxZ (one vs > more than one).
Descriptive statistics of categorical variables focused on frequencies and proportions. Means, medians and ranges were reported for continuously coded variables. The chi-square trend test and t-test were used to compare the statistical significance of differences in proportions, and means, respectively.
Two separate logistic regression models were developed to test the relationship between the number of PBxZ at biopsy and the presence of two endpoints: (i) pathologically unfavourable tumour, defined as a pathological Gleason score ≥4+3 and/or a pathological stage ≥T3, and (ii) clinically significant tumour, defined as a tumour size of ≥2.5 mL [3, 17]. Two separate logistic regression models were developed to test the relationship between tumour laterality at biopsy and the presence of the two previously defined endpoints, namely pathologically unfavourable tumour, and clinically significant tumour. In all four models, covariates consisted of age at diagnosis, PSA concentration, prostate volume, clinical stage, number of positive cores, and number of total cores. In a sensitivity analysis, we restricted our population to patients with extended biopsy scheme (10–19 cores) and repeated the aforementioned logistic regression models in this sub-cohort of patients.
Finally, four patient sub-groups were created, each sub-group corresponded to one of the aforementioned AS criteria [6-9]. All analyses were repeated for each sub-group. All statistical analyses were performed using R statistical package system (R Foundation for Statistical Computing, Vienna, Austria), with a two-sided significance level set at P < 0.05.
Among the 524 patients included in this study cohort, 41.6, 51.3, 75.2 and 95.6% fulfilled respectively the JH, PRIAS, MSKCC and UCSF criteria (Table 1) [6-9]. The pathologically unfavourable tumour rate was 8.4%, ranging from 6.8% (JH criteria) to 8.4% (UCSF criteria; chi-square trend P = 0.5). Clinically significant tumour rate was 25.0%, ranging from 22.0% (JH criteria) to 25.1% (MSKCC criteria; chi-square trend P = 0.5). Other descriptive characteristics are summarised in Table 1.
|Variables||Overall, 524 (100%)||Pathologically favourable tumour, 480 (91.6%)||Pathologically unfavourable tumour, 44 (8.4%)||P||Clinically insignificant tumour, 393 (75%)||Clinically significant tumour, 131 (25%)||P|
|Mean (median, range):|
|Age, years||64.5 (65.4, 47.4–78.6)||64.4 (65.2, 47.4–78.6)||65.6 (66.4, 50.1–78.5)||0.2||64 (65.2, 47.4–78.6)||65.9 (66, 50.1–77.2)||0.002|
|PSA concentration, ng/mL||5.8 (5.7, 0.4–15.0)||5.8 (5.6, 0.4–15.0)||6.2 (6.5, 1.5–9.8)||0.1||5.6 (5.5, 0.4–15.0)||6.4 (6.2, 1.2–12.7)||<0.001|
|Prostate volume, mL||62 (55, 13–190)||62.9 (56, 13–190)||51.2 (50, 20–93)||<0.001||61.3 (56, 16–190)||64 (54, 13–180)||0.4|
|PSAD, ng/mL/g||0.11 (0.09, 0.02–0.45)||0.10 (0.09, 0.02–0.45)||0.13 (0.11, 0.05–0.34)||0.001||0.10 (0.09, 0.02–0.31)||0.12 (0.10, 0.02–0.45)||<0.001|
|Tumour size, mL||1.4 (0.6, 0.005–18.9)||1.3 (0.6, 0.005–18.9)||1.9 (1.2, 0.07–13.2)||0.2||0.3 (0.04, 0.005–2.0)||4.8 (3.8, 2.1–18.9)||–|
|T1c||462 (88.2)||424 (88.3)||38 (86.4)||341 (86.8)||121 (92.4)|
|T2||62 (11.8)||56 (11.7)||6 (13.6)||52 (13.2)||10 (7.6)|
|Total number of cores:||0.9||0.001|
|10–19 cores (extended)||307 (58.6)||281 (58.5)||26 (59.1)||214 (54.5)||93 (71)|
|≥20 cores (saturation)||217 (41.4)||199 (41.5)||18 (40.9)||179 (45.5)||38 (29)|
|Number of positive cores:||0.6||0.5|
|1||192 (36.6)||177 (36.9)||15 (34.1)||146 (37.2)||46 (35.1)|
|2||111 (21.2)||104 (21.7)||7 (15.9)||78 (19.8)||33 (25.2)|
|3||105 (20)||94 (19.6)||11 (25)||82 (20.9)||23 (17.6)|
|≥4||116 (22.1)||105 (21.9)||11 (25)||87 (22.1)||29 (22.1)|
|One zone||393 (75)||366 (76.2)||27 (61.4)||302 (76.8)||91 (69.5)|
|More than one zone||131 (25)||114 (23.8)||17 (38.6)||91 (23.2)||40 (30.5)|
|Tumour laterality at biopsy:||0.009||0.08|
|Unilateral||462 (88.2)||429 (89.4)||33 (75)||352 (89.6)||110 (84)|
|Bilateral||62 (11.8)||51 (10.6)||11 (25)||41 (10.4)||21 (16)|
|Pathological Gleason score:||–||0.001|
|≤6||398 (76)||385 (80.2)||13 (29.5)||314 (79.9)||84 (64.1)|
|3+4||105 (20)||95 (19.8)||10 (22.7)||64 (16.3)||41 (31.3)|
|4+3||18 (3.4)||0 (0)||18 (40.9)||12 (3.1)||6 (4.6)|
|8–10||3 (0.6)||0 (0)||3 (6.8)||3 (0.8)||0 (0)|
|T2||498 (95)||480 (100)||18 (40.9)||378 (96.2)||120 (91.6)|
|T3||26 (5)||0 (0)||26 (59.1)||15 (3.8)||11 (8.4)|
|JH criteria :||0.3||0.2|
|No||306 (58.4)||277 (57.7)||29 (65.9)||223 (56.7)||83 (63.4)|
|Yes||218 (41.6)||203 (42.3)||15 (34.1)||170 (43.3)||48 (36.6)|
|PRIAS criteria :||0.7||0.7|
|No||255 (48.7)||232 (48.3)||23 (52.3)||189 (48.1)||66 (50.4)|
|Yes||269 (51.3)||248 (51.7)||21 (47.7)||204 (51.9)||65 (49.6)|
|MSKCC criteria ||0.8||1.0|
|No||130 (24.8)||119 (24.8)||11 (25)||98 (24.9)||32 (24.4)|
|Yes||394 (75.2)||361 (75.2)||33 (75)||295 (75.1)||99 (75.6)|
|UCSF criteria :||0.157||0.3|
|No||23 (4.4)||21 (4.4)||2 (4.5)||15 (3.8)||8 (6.1)|
|Yes||501 (95.6)||459 (95.6)||42 (95.5)||378 (96.2)||123 (93.9)|
Patients with pathologically unfavourable tumour had smaller prostate volumes (mean 51.2 vs 62.9 mL), higher PSAD (mean 0.13 vs 0.10 ng/mL/g), more frequently harboured more than one PBxZ (38.6 vs 23.8%), and bilateral tumour at biopsy (25 vs 10.6%) than with patients with pathologically favourable tumours (all P ≤ 0.04). Albeit not always statistically significant, these trends were confirmed in patients who fulfilled at least one of the four examined AS criteria (Table 2) [6-9]. There were no statistically significant differences according to pathologically unfavourable tumour status for the other examined clinical characteristics (Table 1).
|Pathologically favourable tumour, n (%)||Pathologically unfavourable tumour, n (%)||P||Clinically insignificant tumour, n (%)||Clinically significant tumour, n (%)||P|
|Entire cohort (n = 524)||PBxZ:||0.04||0.1|
|one||366 (76.2)||27 (61.4)||302 (76.8)||91 (69.5)|
|more than one||114 (23.8)||17 (38.6)||91 (23.2)||40 (30.5)|
|Tumour laterality at biopsy:||0.009||0.08|
|unilateral||429 (89.4)||33 (75.0)||352 (89.6)||110 (84.0)|
|bilateral||51 (10.6)||11 (25.0)||41 (10.4)||21 (16.0)|
|JH criteria  (n = 218)||PBxZ:||0.3||0.2|
|one||189 (93.1)||13 (86.7)||159 (93.5)||43 (89.6)|
|more than one||14 (6.9)||2 (13.3)||11 (6.5)||5 (10.4)|
|Tumour laterality at biopsy:||0.1||0.4|
|unilateral||194 (95.6)||13 (86.7)||162 (95.3)||45 (93.8)|
|bilateral||9 (4.4)||2 (13.3)||8 (4.7)||3 (6.2)|
|PRIAS criteria  (n = 269)||PBxZ:||0.1||0.1|
|one||231 (93.1)||18 (85.7)||191 (93.6)||58 (89.2)|
|more than one||17 (6.9)||3 (14.3)||13 (6.4)||7 (10.8)|
|Tumour laterality at biopsy:||0.07||0.1|
|unilateral||238 (96.0)||18 (85.7)||196 (96.1)||60 (92.3)|
|bilateral||10 (4.0)||3 (14.3)||8 (3.9)||5 (7.7)|
|MSKCC criteria  (n = 394)||PBxZ:||0.02||0.02|
|one||330 (91.4)||26 (78.8)||272 (92.2)||84 (84.8)|
|more than one||31 (8.6)||7 (21.2)||23 (7.8)||15 (15.2)|
|Tumour laterality at biopsy:||0.01||0.01|
|unilateral||346 (95.8)||28 (84.8)||285 (96.6)||89 (89.9)|
|bilateral||15 (4.2)||5 (15.2)||10 (3.4)||10 (10.1)|
|UCSF criteria  (n = 501)||PBxZ:||0.03||0.1|
|one||351 (76.5)||26 (61.9)||290 (76.7)||87 (70.7)|
|more than one||108 (23.5)||16 (38.1)||88 (23.3)||36 (29.3)|
|Tumour laterality at biopsy:||0.01||0.1|
|unilateral||410 (89.3)||32 (76.2)||338 (89.4)||104 (84.6)|
|bilateral||49 (10.7)||10 (23.8)||40 (10.6)||19 (15.4)|
Patients with clinically significant tumour were older (mean age 65.9 vs 64.0 years), had a higher PSA concentration and PSAD (mean 6.4 vs 5.6 ng/mL and 0.12 vs 0.10 ng/mL/g, respectively) than their counterparts without clinically significant tumour (all P ≤ 0.002). There were no statistically significant differences according to large tumour status for the other examined clinical characteristics (Tables 1, 2).
Patients with more than one PBxZ represented 24.8, 7.3, 7.4, 9.6, and 24.8% of the entire cohort, JH, PRIAS, MSKCC, and UCSF criteria based sub-groups, respectively. In these individuals, the rate of pathologically unfavourable tumour was 12.5, 12.5, 15, 18.4, and 12.9%, respectively. The rate of clinically significant tumour was 30.5, 31.2, 35.0, 39.5, and 29.0%, respectively.
Patients with bilateral tumour at biopsy represented 11.8, 5.0, 4.8, 5.1, and 11.8% of patients in the entire cohort, JH, PRIAS, MSKCC, and UCSF criteria-based sub-groups, respectively. In these individuals, the rate of pathologically unfavourable tumour was 17.7, 18.2, 23.1, 25.0, and 16.9%, respectively. The rate of clinically significant tumour was 33.9, 27.3, 38.5, 50.0, and 32.2%, respectively.
At multivariable logistic regression analyses predicting pathologically unfavourable tumour, patients with more than one PBxZ had a 3.2-fold higher risk of harbouring pathologically unfavourable tumour than their counterparts with one PBxZ (Fig. 2A, Table S1, P = 0.01). This risk was 2.6-fold in patients with extended biopsy scheme, and it was 5.8-, 7.2-, 3.5- and 3.7-fold in patients who fulfilled respectively the JH, PRIAS, MSKCC, and UCSF criteria only (all P ≤ 0.02, except for the JH criteria: P = 0.1). Patients with bilateral tumour at biopsy had a 3.3-fold higher risk of harbouring pathologically unfavourable tumour than their counterparts with unilateral tumour at biopsy (Fig. 2B, Table S2, P = 0.008). This risk was 3.6-fold in patients with extended biopsy scheme, and it was 7.4-, 8.7-, 4.5- and 3.2-fold in patients who fulfilled respectively the JH, PRIAS, MSKCC, and UCSF criteria only (all P ≤ 0.04).
At multivariable logistic regression analyses predicting clinically significant tumour, patients with more than one PBxZ had a 2.3-fold higher risk of harbouring clinically significant tumour than their counterparts with one PBxZ (Fig. 3A, Table S1, P = 0.01). This risk was 1.8-fold in patients with extended biopsy scheme, and it was 2.0-, 1.9-, 2.4- and 2.1-fold in patients who fulfilled respectively the JH, PRIAS, MSKCC, and UCSF criteria only (all P ≤ 0.04, except for the JH and PRIAS criteria: both P = 0.2). Patients with bilateral tumour at biopsy had a 1.7-fold higher risk of harbouring clinically significant tumour than their counterparts with unilateral tumour at biopsy (Fig. 3B, Table S2, P = 0.04). This risk was 1.5-fold in patients with extended biopsy scheme, and it was 1.6-, 2.3-, 2.9- and 1.9-fold in patients who fulfilled respectively the JH, PRIAS, MSKCC, and UCSF criteria only (all P = 0.03, except for the JH and PRIAS criteria and patients with extended biopsy: both P ≥ 0.5).
The last decade witnessed an increasing interest in AS as a management option for patients with low-risk prostate cancer . In this context, several criteria have been proposed to identify the most appropriate candidates for AS protocols . Ideally, these criteria should identify men harbouring insignificant and/or low-risk organ-confined tumours. These patients may not benefit from active treatment. Currently, all AS criteria are based on prostate biopsy results as well as on PSA and/or its derivatives. However, none of these criteria include the spatial distribution of positive cores at biopsy as a predictor of unfavourable tumour characteristics. To address this void, we tested the hypothesis that involvement of more than one prostatic zone at biopsy and/or presence of bilateral positive cores at extended biopsy are associated with less favourable tumour characteristics in patients fulfilling at least one of four commonly used AS criteria [6-9].
The present cohort consisted of 524 individuals. Of these, 8 and 25% harboured pathologically unfavourable and clinically significant tumour at RP, respectively. Interestingly, positive cores spatial distribution was able to identify men at higher risk of these adverse pathological findings. Specifically, patients with pathologically unfavourable tumour had more frequently had more than one PBxZ (39 vs 24%), and bilateral tumour at biopsy (25 vs 11%) than patients without pathologically unfavourable tumour (both P ≤ 0.04). At multivariable analyses, the probability of pathologically unfavourable tumour was 3.2-fold higher in patients with more than one PBxZ, and 3.3-fold higher in patients with bilateral tumour at biopsy (both P ≤ 0.01). Similar results were achieved when clinically significant tumour (≥2.5 mL) was considered as the endpoint. Specifically, the probability of clinically significant tumour was 2.3-fold higher in patients with more than one PBxZ, and 1.7-fold higher in patients with bilateral tumour at biopsy (both P ≤ 0.04) after adjusting for all known confounders. It is noteworthy that the relationship between clinically significant tumour and the two predictors of interest (namely PBxZ, and tumour laterality) was evident at multivariable analyses only (Table 2). Differences in clinical characteristics at baseline between patients with and without clinically significant tumour tended to obliterate the examined relationship, which could be detected only after adjusting for possible confounders.
The above-mentioned observations hold true when patients were stratified into sub-groups according to the four examined AS criteria [6-9]. However, when the two most stringent AS criteria (namely JH and PRIAS criteria) were examined, the relationship between positive cores spatial distribution (defined by the number of PBxZ, and/or tumour bilaterality at biopsy) and clinically significant tumour failed to reach a statistically significant level. Similarly, the relationship between the number of PBxZ and pathologically unfavourable tumour failed to reach a statistically significant level in patients that fulfilled the most stringent AS criteria, namely the JH criteria. This might indicate that the positive cores spatial distribution is not a predictor of clinically significant tumour and/or pathologically unfavourable tumour in patients that fulfil the most stringent AS criteria, which include only patients with very-low-risk tumour characteristics. However, it may also be argued that the relatively few patients (n = 218–269) that were included in these criteria undermined the statistical power of the analyses. Indeed, even in patients that fulfilled the JH (most stringent) criteria, the rate of more than one PBxZ was 13% in patients with pathologically unfavourable tumour vs only 7% in patients with pathologically favourable tumour. Likewise, the rate of bilateral tumour was 13% in patients with pathologically unfavourable tumour vs only 4% in patients with pathologically favourable tumour. Similar trends were observed, when clinically significant tumour was examined as an endpoint (Table 2). However, these trends did not reach a statistically significant level. Consequently, these observations warrant further investigation in future studies that benefit from a bigger sample size.
The present results are in contrast with the recent findings reported by Ploussard et al. . They examined data of 300 potential AS candidates (PSA ≤10 ng/mL, clinical stage T1c disease, Gleason score ≤6, <3 positive cores, an extent of cancer in any core 50%, and a life-expectancy >10 years) treated with RP. In their study, tumour upgrading was seen less frequently in cases of a positive biopsy in the midline zone, as compared with positive biopsy at the base, apex, or transition zone. This denotes a relationship between the spatial distribution of positive cores at biopsy and pathological tumour characteristics. However, they failed to show a statistically significant relationship between unfavourable tumour characteristics and the number of PBxZ, and/or tumour bilaterality at biopsy. This lack of association may be attributed to the relatively small cohort (n = 300) studied by Ploussard et al.  and/or to the very stringent AS criteria examined in their report. To the best of our knowledge, no other reports have focused on a similar subject.
The present results offer a new perspective, because they show the importance of positive cores spatial distribution in selecting AS candidates. It appears that among AS candidates, one out of 10 patients harbour a pathologically unfavourable tumour, and one out of four patients harbour clinically significant tumour. These rates were virtually equal for all the four examined AS criteria. The inclusion of these individuals with misclassified tumour in AS protocols may compromise their cancer control outcomes. Such numbers might become clinically relevant due to the increasing use of AS among patients with low-risk prostate cancer. To reduce such rates of misclassification, the spatial distribution of positive biopsy cores should be used. For example, the use of tumour bilaterality as an exclusion criterion in AS protocols would preclude AS in 5–12% of the individuals. However, 17–25% of these individuals have pathologically unfavourable tumour, and 27–50% have clinically significant tumour. Virtually all the available current AS inclusion criteria might benefit from the use of at least one of the two indicators of spatial distribution of positive biopsy cores, namely the number of PBxZ and tumour laterality at biopsy (Table 2). Similarly, these indicators might also be used in the follow-up biopsies scheduled for patients managed with AS to trigger active treatment. This would potentially identify patients at a higher risk of progression and may reduce the likelihood of inappropriate surveillance even during the follow-up period. However, this hypothesis still needs to be tested in future studies.
Despite these strengths, the present study is not devoid of limitations. First, we focused on pathologically unfavourable tumour and clinically significant tumour as endpoints. However, these two endpoints may not necessarily translate into less favourable cancer control outcomes. This limitation is shared with several previous reports that addressed similar endpoints [20-23]. Second, the present study also spanned a long period, and the grading of biopsy and RP specimens changed over time . Third, there were variations of biopsy scheme over time. However, we accounted for the total number of cores as well as for the number of positive cores in multivariable analyses, thus accounting for the effect of such variations. Moreover, in a sensitivity analysis that focused exclusively on patients with extended biopsy schemes, our observations remained virtually the same. It is also noteworthy that the total number of cores (and the number of positive cores) in each biopsy zone (right/left apex, right/left margin, right/left base, and transition zone) varied according to the biopsy scheme that was used. However, the total number of biopsy zones (six peripheral zones + transition zone) remained stable over the study period. Consequently, it is unlikely that changes in biopsy scheme over the study period have significantly influenced our findings. Fourth, although the collection of data was performed prospectively, the retrospective design of the study may be considered a limitation. Finally, the present cohort represents single institution data where no AS protocols are currently ongoing. It remains to be tested whether our findings are applicable to other clinical settings or countries where PSA screening programmes are implemented more.
In conclusion, the spatial distribution of positive cores at biopsy should be considered, when advising patients about AS. Specifically, patients with more than one PBxZ and/or bilateral positive cores at biopsy should be considered at higher risk of harbouring a pathologically unfavourable tumour and/or clinically significant tumour. Consequently, AS might be associated with less favourable cancer control outcomes in these individuals that might benefit from active treatment. Inclusion of spatial distribution of positive cores in AS criteria might decrease the rate of prostate cancer misclassification, thus optimising cancer control outcomes in these patients. However, the most stringent AS criteria (namely JH and PRIAS criteria) might not benefit from the addition of this predictor. This point warrants further investigation in future studies.
This research received no specific grant from any funding agency in the public, commercial, or not-for-profit sectors. Authors do not have other financial disclosure or conflict of interest to declare.