The main finding coming from the analysis of this sub-cohort was the duration of prostate atrophy after the single injection of 200U of BoNT/A which was found to be about 18 months. As previously stressed, the duration of this period had never been reported. In addition we found that the nadir of prostate volume occurred 6 months after the neurotoxin administration.
The decrease of prostate volume uniformly described after intraprostatic BoNT/A injections [2, 10, 12, 13] should be related with the widespread apoptosis that has been detected in the gland after BoNT/A administration in rats, dogs, and humans. Other mechanisms such as a straight necrotic effect of the injected solution seems improbable as cavitations were not observed in the gland tissue during transrectal ultrasounds repeatedly carried out in this or in other studies [2–10, 12, 13]. Apoptosis involves both the epithelial and stromal components [3, 8, 14] and recent observations suggest a mechanism of action involving the impairment of the autonomic innervation . As a matter of fact, Silva et al, demonstrated in the Rat that gland apoptosis induced by intraprostatic injection of BoNT/A was caused by the lack of a continuous sympathetic drive . BoNT/A-induced apoptosis could be prevented by maintaining the adrenergic stimulation of the gland with phenylephrine, an adrenergic agent, administered subcutaneously . In spite of the rich cholinergic innervation previously reported in the prostate  the impairment of the parasympathetic innervation to the prostate atrophy was less remarkable. Cholinergic stimulation could not prevent BoNT/A-induced atrophy .
Despite the marked prostate atrophy, total serum PSA had a very moderate decrease, not exceeding 25% at 3–6 months after BoNT/A. The changes in serum PSA after BoNT/A injection are now at the centre of a lively debate. One would expect that a decrease of total serum PSA would accompany prostate tissue decrease, as usually observed after prostatectomy. Moreover, Maria et al [2, 13] and Guercini et al [6, 13], reported decreases over 50% in serum PSA levels after prostate volume decreases. It should, however, be stressed that other studies, like us, did not find any decrease in total serum PSA levels regardless of a significant prostate atrophy. This is the case of Chuang et al [8, 13] and Park et al [10, 13] in which changes of total serum PSA were non-significant. Therefore, at this moment it is unclear why total serum PSA and prostate size volume may not walk side by side after BoNT/A injection. Nonetheless difficulties in matching prostate size and PSA variation have also been observed in other studies. Loeb et al  stated that prostate volume increases, determined by MRI, and PSA velocity do not correlate. Despite growth rates as high as 10 cc per year, PSA velocity was less than 0.1 ng/ml per year in most men without prostate cancer . Likewise, prostate size and PSA decreases observed after 5α-reductase inhibitors administration never attained similar values, the decrease of serum PSA reaching 50% whereas prostate volume not exceeding 25% .
Resumption of spontaneous voidings by the patients included in this group should not be attributed solely to prostate volume reduction, as a decrease in prostate tonus was also reported after BoNT/A administration . Lin et al observed that injection of 200U BoNT/A significantly reduced prostate urethral pressure in response to intravenous adrenergic agents or electric stimulation in dog prostates. Interestingly, prostate tonus, is under sympathetic control , the autonomic innervation that is severely impaired by intraprostatic BoNT/A .
As all patients were in indwelling catheterization, PVR, IPSS score and Qmax were evaluated only after the successful voiding that occurred when the urethral catheter was removed 1 month after BoNT/A injection. This moment to remove the catheter was chosen empirically as such a study had never been conducted before. Nevertheless we took in consideration that Maria and coworkers  had reported a marked reduction in prostate volume and a discernible improvement in peak urinary flow 1 month after BoNT/A injection, suggesting that at this time point beneficial effects of the neurotoxin were already present. Average IPSS score varied between 10 and 12, clearly indicating that LUTS severity remained low  once the catheter was removed and spontaneous voidings were resumed. In accordance, a low score of the associated QoL question was found. In what concerns flow, the mean values ranged between 10,5 and 12,3 ml/sec. Although it might be argued that these Qmax values are not exceptionally high, it should be remembered that at baseline all these 11 patients were on chronic indwelling catheterization refractory to several voiding trials. In addition, PVR remained well below 100 ml during follow up indicating that bladder emptying was rather effective.
Intraprostatic BoNT/A injection was carried out transrectally. This route was used for the first time in our initial study  and it was preferred due to the experience the authors accumulated while obtaining prostate biopsies oriented by transrectal ultrasound. In addition, our preliminary attempts revealed that this route is more reproducible than the perineal one described by Maria et al  and carries less risk of morbidity, namely hematuria, than the transurethral route used by Kuo et al . Transrectal injections were well tolerated as attested in this cohort by the lack of local or systemic side effects. The fact that general anaesthesia was not required for prostate injection was also a clear advantage as it allowed the procedure in the frail elderly patients included in this study, who had a mean age above eighty years and significant comorbidities. All the patients went home immediately after the treatment which was performed as an office procedure.
One should realise that the study has several limitations. The first comes from the fact that the number of patients that could be followed during 18 months is small and represents a sub-cohort of patients that responded better to BoNT/A injection. However the other 10 patients of the initial cohort that could not be followed for the reasons pointed above, all had a marked atrophy at six months evaluation . Therefore the patients reported in the present study do not represent a special subgroup with prostate atrophy but rather a subgroup with better clinical outcome and/or better clinical condition to attend regular visits. Another limitation comes from the fact that no control group exists, making a bias towards prostate atrophy possible, albeit improbable taking in consideration the magnitude of absolute changes. Pressure/flow studies were not carried out, preventing therefore the correct determination of the cause of the refractory urinary retention. However one must understand that in such a group of frail elderly patients with large prostate glands, urinary retention due to bladder prostatic obstruction is the likely diagnosis. Although bladder decompensation is a well known problem in this aged group of patients, a hypotonic bladder should not be the most important reason for the refractory urinary retention present at inclusion. In fact, these patients voided spontaneously after prostate volume/prostate tonus reduction brought on by BoNT/A. The third limitation is the lack of IPSS scores and QOL scores at baseline, as patients did not void spontaneously at inclusion. Thus, IPSS score, Qmax and PVR determined along the study should be interpreted as an indication that the BoNT/A effect persisted also during a long period of time. Finally, it is unclear if the same time-response will occur if small prostate glands are injected with BoNT/A.