The major objectives of this single arm, multicenter, phase 2 study were to evaluate kidney angiomyolipoma response to and tolerability of sirolimus treatment. Overall, we found that sirolimus is relatively safe and may be a useful systemic option for treating patients with problematic multifocal kidney angiomyolipomas and other TSC related tumors. We observed that responses persisted at 24 months in a subset treated with additional sirolimus during year 2 of the study. We also observed regression of liver angiomyolipomas and SEGAs. Furthermore, serum VEGF-D levels were elevated at baseline, decreased with sirolimus treatment, and correlated with kidney angiomyolipoma size (for a comparison of our VEGF-D results to several sporadic LAM studies, see Table S4
There are two other published kidney angiomyolipoma clinical trial studies so we compared the kidney angiomyolipoma response rate and toxicity data from our study (36 enrolled, 28 completed 1 year of sirolimus treatment) with the two related studies 
. One is a single institution sirolimus study (25 enrolled, 20 completed 1 year of sirolimus treatment) and the other is a UK multicenter sirolimus trial (16 enrolled, 10 completed 2 years of sirolimus treatment). Both our study and the UK study used RECIST criteria. We observed an overall response rate of 16/36 (44.4%) and the mean decrease in the sum LD for kidney tumors size was 29.9% with 1 year of treatment. In the UK study, the response rate was 8/16 (50%) with 2 years of treatment, and the mean decrease in the sum LD for kidney tumors was 25% at the 1 year time point. The single institution study used volumetric data to evaluate tumor regression and found that the mean reduction in volume was 53.2% with 1 year of treatment. Since RECIST uses a one dimensional measurement (sum LD) and volume is a three dimensional measurement, it is important to note that a 25–29% decrease by RECIST corresponds to a ~59–64% decrease in volume if we assume tumors are approximately spherical. All three studies showed that kidney tumor regression was observed in all study subjects who completed 1 year of treatment, although in some cases tumor regression did not meet criteria for response by RECIST (which requires ≥30% decrease in sum LD). Kidney tumor regrowth was observed at 24 months in the subset of participants in this study who were followed off therapy after week 52 (n
15) and in most of the single institution cohort (n
18, all of whom were followed off therapy after week 52). Kidney tumor responses tended to persist in a subset of our study participants who were treated with additional sirolimus from months 12 to 24 (n
13), and in all of the UK study participants who completed 24 months on study (n
10). There were also many similarities regarding toxicity findings across all three studies. Serious drug related events were relatively rare (3 in our study, 6 in the single institution study, 3 in the UK study). However, mild and moderate toxicities were common and included mouth ulcers, hyperlipidemia, infections, proteinuria, and diarrhea. Additionally, our SEGA results are similar to the SEGA regression observed in a recently published single institution study of another mTOR inhibitor (everolimus) that showed a mean SEGA volume decrease of 49% at 12 months and 57% at 24 months) 
Progressive LAM is a devastating disorder with a known defect in the mTOR pathway 
. Both published kidney angiomyolipoma sirolimus treatment trials also collected pulmonary data so we compared our PFT results to those studies. Looking at the data from individual subjects, in our study we found that after 52 weeks of sirolimus treatment, FVC increased by more than 5% in 5/15 (33%) TSC/LAM subjects (3 moderate LAM, 2 mild LAM) and FEV1 increased by more than 5% in 4/15 (27%) TSC/LAM subjects (2 moderate LAM, 2 mild LAM), see Tables S7
and Figures S5
. In the UK multicenter study, FVC decreased by 76±52 ml/year and FEV1 decreased by 55±94 ml/year (n
5 subjects with PFT results over 2 years). The single institution study reported week 52 PFT data on 11 cases with LAM (5 sporadic LAM and 6 with TSC/LAM) and noted a higher frequency of improvement in PFTs. In that study, FVC increased by more than 5% in 8/11 (73%) and FEV1 increased by more than 5% in 5/11 (45%). A major limitation in all of these studies is that the number of patients included is low, there is no untreated control group, pulmonary function testing is effort dependent, and there were differences in the LAM population included in the three studies. Furthermore, it appears that the single institution study may have included subjects with more severe lung disease as the average FEV1 and FVC at baseline were lower in their cohort than in ours and the UK study. Since we observed a tendency for improvement in the more severely affected group, this could be one explanation for the more dramatic findings in the single institution study. More recently, a randomized phase 3 trial of sirolimus for the treatment of LAM showed that the sirolimus group had significant improvement in FEV1 and FVC after 1 year of sirolimus treatment (n
46 enrolled) compared to the placebo group (n
43 enrolled). In this phase 3 study, subjects with more severe LAM were enrolled (FEV1 of <70% of predicted was required for study entry) and mean FEV1 remained stable in the sirolimus group (change from baseline to 12 months was 19±24 ml) compared with the placebo group (−134±182 ml, p
<0.001). Additionally, mean FVC showed some improvement in the sirolimus group (97±260 ml) compared with the placebo group (−129±233 ml, p
. These numbers are quite similar to our findings that in subjects with LAM (n
15), the mean FEV1 change from baseline to week 52 was 20 ml and the mean FVC change from baseline to week 52 was 130 ml. A limitation of all LAM studies completed to date is that the treatment period was relatively short (12 months) so longer term studies are now needed. Although several studies now show improved pulmonary function with sirolimus treatment, it is not clear that that long term benefits will outweigh risks, especially considering that sirolimus is an immunosuppressant and pneumonitis is a known and potentially serious toxicity of mTOR inhibitors 
. Ideally the benefits and risks of using of sirolimus for the treatment of LAM will be further evaluated in longer term studies.
To our knowledge, this is the first clinical trial to report VEGF-D results in a TSC population with kidney angiomyolipomas before and after mTOR inhibitor treatment. The correlation between VEGF-D and kidney tumor size suggests that serum VEGF-D levels may be useful for monitoring kidney angiomyolipoma size over time. Our observations that VEGF-D levels are elevated at baseline and decrease with sirolimus treatment in a population with kidney angiomyolipomas associated with kidney disease are consistent with the VEGF-D results reported in the phase 3 sirolimus trial for women with LAM reported recently 
. The clinical application of using VEGF-D levels to monitor kidney angiomyolipomas or other features of TSC and/or LAM will need confirmation in future studies. The positive results indicate that the inclusion of VEGF-D and other biomarker studies in future trials may yield new tools for clinical management and contribute to the understanding of disease pathology for TSC and/or LAM.
The numerous clinical features of TSC are well known 
and well represented in our study participants (). There are significant unmet medical needs in this population; involvement of multiple organs is common, and many with TSC are living with chronic disabilities. The data reported here indicate that sirolimus treatment is a promising new approach to treatment for problematic kidney angiomyolipomas, liver angiomyolipomas, and SEGAs. Strengths of this study include: it is a multicenter study so is less prone to selection bias than single institution studies, tumor response was observed in multiple organ systems (kidney, brain, liver, skin) and this is the first TSC trial to identify a potentially useful biomarker (VEGF-D) for kidney angiomyolipomas. Limitations of this study include: it is a single arm, open label study, the follow-up duration was only 2 years, pediatric subjects were not included, and skin lesion assessment was subjective. Future trials for TSC related tumors should include longer duration sirolimus studies and pediatric subjects. Because the severity of kidney angiomyolipomas and other manifestations of TSC are variable, randomized trials that include patient reported or quality of life outcomes will have the highest impact regarding guiding the optimal use of mTOR inhibitors for the treatment of TSC. There is preclinical 
and case report data 
showing that lower sirolimus doses may be equally effective in treating TSC related kidney tumors. Therefore, trials using lower maintenance doses to see if the response is preserved while minimizing toxicity should also be considered. Brain manifestations (seizures, cognitive impairment, and behavioral problems) are a major source of morbidity for individuals with TSC. The improvement in cognition and seizures with sirolimus in mouse models 
, together with the efficacy and safety data reported here and other TSC mTOR inhibitor trials, indicate that clinical trials in children and adults with more severe manifestations of these features are now justified. Our skin lesion data, together with preclinical and case report data 
indicate that studies evaluating systemic and/or topical sirolimus for treating TSC skin lesions are worthwhile. Our pulmonary function data suggest that longer term studies of pulmonary outcomes in women with progressive and/or moderate to severe LAM are of interest. As there are now several ongoing TSC and/or LAM trials investigating the utility of RAD001 (a rapamycin analog, also known as everolimus and Afinitor), it would also be useful to pursue comparative effectiveness studies in order to directly compare efficacy, safety, and costs of different mTOR inhibitors.
There are recent preclinical studies using mouse models of TSC related tumors that indicate other approved drugs (such as angiogenesis inhibitors, interferon-gamma, asparaginase) may have potential therapeutic utility as single agents or in combination with mTOR inhibitors 
. However, these studies show that non-mTOR inhibitors appear to be less effective than single agent mTOR inhibitors. Based on this, we suggest that clinical investigation of non-mTOR inhibitors as single agents or in combination with a mTOR inhibitors should be evaluated as second line therapy for problematic TSC related tumors that are not responding to an mTOR inhibitor. As the response of TSC related tumors in mouse models is consistent with response of TSC related tumors in humans, we anticipate that future preclinical studies have significant potential to identify other promising new medical approaches for the treatment of TSC and/or LAM.
This multicenter study should provide useful efficacy and safety data to clinicians who are considering the option of recommending sirolimus treatment for individuals with TSC and problematic kidney angiomyolipomas or other TSC associated tumors. Although there are alternative treatment options for problematic kidney angiomyolipomas (vascular embolization, nephrectomy, partial nephrectomy), these are all local, invasive procedures with associated risks. Furthermore, the available local interventions are often not suitable for the bilateral, multifocal kidney tumors that are frequently observed associated with TSC 
. Ultimately it will be important to demonstrate that mTOR inhibitors can improve longer term outcomes in TSC and/or LAM patients with problematic kidney angiomyolipomas. Although there are challenges that slow the rate of clinical research progress for uncommon disorders 
, the quickest path to evaluating the benefits and risks of long term treatment is to pursue evaluation of mTOR inhibitors in larger numbers of affected individuals in the context of clinical studies that include substantial long term follow-up for both efficacy and safety endpoints.