Tuberous sclerosis is a heterogeneous disorder that typically presents during infancy. Our research revealed that the median age of TSC diagnosis was 8 months, aligning with the multi-country TOSCA study involving 2211 TSC patients spanning all ages [18]. TSC2 mutations were more prevalent than TSC1 mutations, with a ratio of 3.7 to 1, in line with previous studies [18, 19]. However, it should be acknowledged that mutation analysis was not routinely conducted in 26.4% (48/182) of children in our study. Nair and colleagues [20] highlighted that approximately 70% of TSC cases arise from de novo mutations; our study found 80.2%. Our study demonstrated higher organ involvement in patients with TSC2 mutations than TSC1 mutations, consistent with earlier findings that TSC2 is associated with more severe disease characteristics [21]. Back pain attributed to kidney lesions in TSC patients is an uncommon presentation and has rarely been reported in previous studies, but our study identified five children (2.7%) who presented with back pain. All the cases were in children above 8 years of age and had nephromegaly, though they presented with varying kidney phenotypes.

Detecting kidney involvement in TSC necessitates imaging to identify lesions, as small lesions may be incidental findings, and most patients with TSC kidney disease are asymptomatic. MRI is the preferred imaging modality for diagnosis and follow-up of TSC-related kidney lesions stemming from its superior soft tissue resolution and multiplanar capabilities without radiation exposure [22]; however, despite its recommendation in the latest consensus [22, 23], MRI was used in only 23.5% of cases in our study. The limited utilisation of MRI in our cohort can be attributed to several factors. MRI is primarily chosen for the evaluation of large AMLs to monitor growth and complications and MRI often necessitates sedation in younger children and may be avoided by caregivers of children with TSC [22]. An alternative imaging modality is ultrasound, accounting for over 75% of the imaging used in our study. However, it should be noted that the quality of ultrasound imaging is operator-dependent and may miss sizable lipid-poor masses (such as AMLs) and echogenic kidney foci that are too small to characterise into either AMLs or cysts [11, 24]. Moreover, conducting ultrasound scans on children with TSC can be challenging. The frequency of kidney assessment should be individualised, with annual follow-ups deemed sufficient for stable or lesion-free patients [23, 25]. Considering differences in disease progression between TSC phenotypes, we recommend genetic testing to be performed in all patients with suspicion of TSC, and more frequent monitoring of kidney function and imaging (at least annually) is advised for older children, with an earlier age threshold for children with TSC2 mutation. Current recommendations advocate for CT use only if MRI is unavailable, with CT angiography reserved for AMLs above 3 cm to exclude intra-lesional microaneurysms or pinpoint bleeding sources in haemorrhagic cases, despite weak evidence supporting this recommendation [8, 25]. However, no CT scans were conducted in our study due to concerns about radiation risk.

Among the 145 examined children, kidney imaging revealed lesions in 114 (78.6%), comprising cysts (53.8%), AMLs (61.1%), or both (45.6%). This concurs with Bissler et al.’s [5] recent review on kidney involvement. While kidney cysts were slightly more prevalent in males and AMLs in females, we did not find a significant association between gender and lesion prevalence, consistent with other reports [26, 27]. The TSC genotype is a predictor for kidney involvement [20], as supported by our results showing higher incidences of both cysts and AMLs in patients with TSC2 mutations compared to TSC1; however, this significance was noted only for AMLs, not for cysts. Our results, echoing Kingswood et al. [18], highlighted significantly higher rates of bilateral AMLs in TSC2 patients than in TSC1 patients. This finding aligns with data indicating multiple and bilateral distributions of most AMLs and cysts [11, 26]. Interestingly, we observed that the largest AMLs were primarily in the left kidney (72.7%). Kidney lesion laterality has not been investigated extensively, but previous studies on RCC have also found that patients with left-sided lesions had poorer outcomes than those with right-sided RCC [28], suggesting that lesion laterality in kidneys may not be merely coincidental. Kidneys differ in anatomy, vascular supply, and lymphatic drainage between the left and right kidneys; the left kidney has more vascular collateral circulation and lymph nodes. Cutaneous TSC lesions demonstrated proliferation and dilatation of lymphatic and blood vessels on histology [29]; as such, this increased vascularity may deliver more vascular endothelial growth factors necessary for angiogenesis and lymphangiogenesis, inducing the development of larger AMLs found in the left kidney.

Cook et al. [30] suggested that cysts were more common than AMLs in children under 5. Our study found that over 40% of children under 5 had cysts, compared to approximately 25% with AMLs. Note that we were unable to exclude the possibility that patients may have already had kidney cysts or AMLs prior to the initial imaging episode. Kidney AMLs are relatively infrequent in children under 2, likely due to their slow growth and small size, often missed on imaging. Our data show that the change in the size of the largest AML accelerates with age: 1.03 mm/year (ages 5–9 years), 2.29 mm/year (ages 9–14 years), and 2.82 mm/year (ages 15–19 years). Robert and colleagues [31] also observed AML growth doubling before and after 12 years (from 2 mm/year to 4.5 mm/year), though their observation was based on data from 21 patients with AML, and the authors did not specify how these rates were obtained. Interestingly, Ewalt et al. [32] reported a case of an 18-year-old man with AML growth at 4 cm per year, although they did not provide further details about this case, including the genetic makeup. The influence of oestrogen and progesterone may explain this surge, given the hormonal receptors on the tumour surface [33]. As such, peri-pubertal, pregnancy, or hormonal treatment phases could significantly boost AML growth due to the enhancing effects of these hormones. Also, while ultrasound may be too insensitive for accurate measurements of kidney lesions, MRI also faces limitations as the precision in measurements depends on its slice periodicity (i.e. growth less than a single slice distance), and thus inaccuracy in measuring smaller lesions is possible. Prospective studies involving expert radiologists are necessary to accurately measure and calculate the longitudinal growth rates of individual lesions, aiding in determining the recommended interval for follow-up.

To the best of our knowledge, only a Belgian study, conducted by Janssens et al., has explored cyst growth in TSC cases [7]. They observed a median cyst growth of 0.2 mm/year among 45 patients; however, the researchers did not conduct any further analysis based on age or genetic mutations. Our study yielded novel findings, with kidney cysts displaying two peaks: the first in children under five and another between 15 and 19 years. We acknowledge that our findings might have been confounded by including four patients with CGS, where CGS patients are known to experience more severe polycystic kidney growth and earlier onset of kidney impairment [8]. Even after the exclusion of these four cases from our analysis, a similar pattern persisted. Although we could not find existing literature explaining these specific findings, we hypothesise that a combination of factors might contribute to increased cyst growth during the first 5 years of life. These factors include heightened cellular proliferation, a rapid kidney growth rate (especially in the early years) [34], and, to a lesser extent, an altered hormonal milieu in early childhood [35].

In our analysis, we observed variations in the time to cyst development between children with TSC1 and TSC2 mutations, albeit the difference did not reach statistical significance, likely due to the study being underpowered to detect kidney cysts from the limited TSC1 cases (n = 20). Nevertheless, TSC1 demonstrated better cyst-free survival (median 16.9 years) than TSC2 (median 9.1 years) at any time. On another note, our study underscores the importance of conducting dedicated studies focused solely on children with TSC. For instance, the TOSCA study that included patients across all ages revealed the mean age at AML diagnosis of 22.5 years for TSC1 and 13.3 years for TSC2 [18]. In contrast, our study observed a median age at AML development of 12.7 and 7.8 years, respectively. These disparities in the age of kidney lesion onset highlight the imperative of acknowledging such variations when formulating clinical practice guidelines for managing these cases.

While most patients with AMLs remain asymptomatic [25], those that exceed 3 cm in size are associated with an increased risk of haemorrhage—the main complication of AML and a leading cause of mortality in TSC [27, 36]. In our study, we observed that 9% (13/145) of children developed AMLs above 3 cm, with the majority exhibiting TSC2 mutations. The median age at which AML size exceeds 3 cm was 13.8 years (range: 5.3–18.9 years). Subsequent follow-up primarily utilised MRI. Notably, more than half of these patients (7/13) experienced AML growth beyond 4 cm, with only four of them receiving everolimus treatment. Current clinical guidelines suggest that individuals with AMLs exceeding 3 cm may benefit from everolimus treatment, as shown to hold therapeutic potential in the EXIST-1 to EXIST-3 trials [22, 23, 36, 37]. However, other factors, including physician expertise and clinical judgement, may influence the decision to initiate everolimus treatment. Among the remaining nine children receiving everolimus treatment for neurologic indications, none developed high-risk kidney lesions, and we lacked sufficient data to determine the effect of everolimus on AMLs of relatively small size. Nonetheless, none of the patients with high-risk AMLs experienced acute bleeding. Apart from haemorrhage, TSC kidney lesions can progressively replace functional tissue, leading to early GFR impairment and secondary hypertension. Our study observed AMLs exceeding 3 cm were significantly associated with developing nephromegaly (p = 0.028), hyperfiltration (p = 0.014), CKD (p = 0.001), and being initiated on everolimus treatment (p = 0.007) (see Supplementary Table 4).

In our study, three (1.6%) of 182 children with TSC underwent kidney interventions, including partial nephrectomies in 2 children and embolisation in 1. The two cases of partial nephrectomy were performed in children aged 8 and 14 years. In one case nephrectomy was indicated due to nephromegaly (right kidney > 21 cm). Neither hypertension nor other complications were detected pre- and post-nephrectomy in both cases. Only one patient underwent embolisation due to AMLs measuring over 5 cm, without acute bleeding. Our findings are consistent with the TOSCA study, which suggests that everolimus usage is more common than embolisation and nephrectomy combined [12].

Hyperfiltration, hypertension, and proteinuria are known risk factors for kidney disease progression [7]. Although these factors were prevalent in our study, we did not find a significant association with CKD, similar to the findings of the Belgian study [7]. It is worth noting that the definition of glomerular hyperfiltration based on eGFR is still lacking and not validated, and thus may affect its accuracy in capturing the true incidence of hyperfiltration [16]. Out of 81 children, thirty-nine (48.1%) experienced at least one episode of hyperfiltration. Although overestimation is possible due to methodological constraints, and some patients lacked longitudinal kidney function data on record, evidence suggests that hyperfiltration was transient in some children. Furthermore, a third of the cases with a hyperfiltration episode were detected in children under the age of 5 years, supporting the recommendation that at least annual biochemical testing to monitor kidney function should be performed in children with proven kidney involvement on imaging [22]. CKD, which we defined as stages II–V and eGFR below 90 mL/min/1.73 m2, was identified in 10 children in our study: 8 children were at stage II, 1 child at stage III (eGFR < 60), and 1 child at stage IV (eGFR < 30). Both cases with advanced CKD (stage III–V) had TSC2/PKD1 mutation. The child with stage III CKD (aged 5 years) had developed multiple small kidney cysts and multiple AMLs, with two AMLs measuring above 4 cm with multiple antihypertensive drugs. The other case with CKD stage IV (aged 16 years) had an eGFR of 29 and developed bilateral, multiple, and large cysts on both kidneys. No other potential cause of CKD was identified in this case. Mekahli et al. [22] highlighted that the precise proportion of TSC patients with CKD remains unclear, partly due to the absence of TSC as a diagnosis code in kidney failure databases. Regarding other variables analysed, such as hypertension or being on everolimus, no significant associations were observed with the presence of kidney lesions, high-risk AMLs, or even after stratification by age groups. This lack of significant findings is likely attributable to the study being underpowered to determine meaningful predictions.

Limitations of this study include the retrospective nature resulting in varied follow-up times for patients, potentially leading to an underestimation of kidney lesions or complications in some cases. The exclusion of patients without imaging reports may introduce bias, and reliance on reports from multiple radiologists may lead to imprecise measurements and difficulty distinguishing disease progression from previously missed lesions. Missing data in patient records could also distort our findings or result in underpowered analyses; however, our sensitivity analyses revealed significance only for cyst-free survival. Our study did not longitudinally track the serial dimensions of kidney lesions, but instead data were collected in a cross-sectional manner with only the largest lesion noted at each time point. Consequently, this may potentially lead to an underestimation of lesion growth.

In conclusion, this study offers valuable insights from 145 children with tuberous sclerosis and kidney imaging data available, which is a substantial cohort given the rarity of the condition. Novel findings in this study include the location of kidney lesions, kidney cyst growth, and survival analyses for both kidney cysts and AMLs. While TSC kidney disease emerges later in the disease course than neurological features, routine and effective surveillance of the incidence and growth velocity of cysts and AMLs and their progression, including regular kidney imaging, kidney function, and blood pressure monitoring, should be adopted for these patients during childhood.