Using a mathematical model we evaluated clinically relevant questions of BCG therapy in NMIBC. The model revealed that therapeutic success depends strongly on the timing of the BCG regimen. An early start of BCG therapy after TUR, combined with an optimal dwell time and treatment inter-instillation interval, do, in fact, have profound influence on treatment outcome, according to our model.
There are no current guidelines advising when to start BCG therapy after surgery. Delay is considered important for the healing of the bladder wall, and the prevention of systemic complications due to BCG therapy. Therefore, most urologists wait 2–6 weeks prior to starting BCG therapy; however, in some studies intravesical treatment has been initiated as early as 1 week after transurethral resection 
. To date, no prospective comparisons of different delays to start of BCG treatment have been performed. Our model indicates that a prolonged delay in initiating BCG therapy could negatively impact recurrence rates (). Analysis of the model suggested that the increased risk for recurrence was related to outgrowth of residual tumor cells. This result is due to the continued expansion of residual tumor cells post-resection, increasing the burden of disease and challenging the limited kill capacity of the immune system.
Changes in dose and dwell time have been previously discussed in the literature on the basis of reducing side effects. Dose reduction has been assessed in several clinical trials 
. Reducing BCG dose to one third was considered as a strategy aimed at lowering side effects. This lower dose remained significantly better than mitomycin; whereas one sixth of the BCG dose was not better than the use of mitomycin alone 
. These findings corroborate our modeling results regarding the influence of dose on the cure rate (). Our model also shows that increasing the dose increases therapeutic success. However, this would probably occur at the cost of enhanced side effects.
Reduction of dwell time has been reported as a possibility for improving therapy and as an alternative to dose reduction in patients with severe side effects 
but no prospective study has compared BCG dwell time as a variable. Our model indicates that increased dwell time might also influence treatment outcome (). This could be of special importance in patients with minimal symptoms that may benefit from enhanced BCG-mediated immune activation. These results provide a stark reminder of the importance of adhering to, at minimum, the current guidelines for BCG immunotherapy.
Perhaps the most striking finding is the observation that the optimal treatment interval could be twice as long as the current schedule for managing patients (). This finding is likely related to the kinetics of T and B cell activation, and the persistence of these relatively long-lived effector cells in the bladder. By extending the treatment interval during the effector phase, we have succeeded in enhancing the time period in which the immune system may exert negative pressure on the residual tumor burden. Only limited information is available from clinical trials where the treatment interval was modified during BCG induction therapy. Studies in mice indicated that the number and timing of the instillations are important in determining different local cytokine profiles, which in turn may influence the qualitative and quantitative recruitment of adaptive effector cells 
. Such findings in combination with the result of our model support the need for further investigations to determine the optimal timing of BCG instillations. Finally, attention should be paid to the abrupt loss of the treatment effect after a certain interval (), a finding that may have implications for the timing of maintenance therapy.
These data are intriguing as the modified treatment regimen engages the afferent immune response during the early phase of treatment and maximally benefits from the long-lived effector potential of the efferent adaptive response. Such a regimen with extended courses of treatment may be better tolerated by patients 
. Moreover, it is intriguing to consider that extension of the first six doses of BCG could obviate the need for maintenance therapy. Clearly, these findings require validation in pre-clinical experimental models and clinical trials prior to their being adopted for patient management. Nonetheless, we are encouraged by the findings and support the use of mathematical models to establish a framework for optimization of treatment practices.
The field of mathematical modeling in BCG immunotherapy of bladder cancer has emerged only recently 
. Topics addressed so far have been: BCG dose and number of instillations needed to achieve cure, and the combined effect of IL-2 and BCG. Previous models 
have a number of limitations, which have been circumvented in this work. First, instead of ordinary differential equations, where the numbers of cells are given by real numbers, we have chosen a stochastic model where the numbers of cells take integer values. Hence, our model is better suited to describe tumor elimination. Second, unlike previous models 
, ours includes the dynamics of BCG-associated healthy urothelial cells, which are more numerous than BCG-associated tumor cells – both serving as initiators of innate and adaptive immunity. Third, we are the first to model the prime boost response of the immune system, a phenomenon believed to be of critical importance for BCG immunotherapy 
. These methodological advances, in addition to the results presented here, will help advance the use of mathematical models for optimization of immune-based treatment strategies.