Exonic Mutations of TSC2/TSC1 Are Common but Not Seen in All Sporadic Pulmonary Lymphangioleiomyomatosis

To the Editor:

Pulmonary lymphangioleiomyomatosis (LAM) is a rare, progressive disease affecting almost exclusively women that is seen in both patients with tuberous sclerosis complex (TSC) and those without TSC, the latter termed sporadic LAM (S-LAM) (1–5). LAM is characterized by infiltration of the lung parenchyma by neoplastic spindle-shaped cells with combined smooth muscle and melanocytic differentiation, and is associated with both extensive involvement of lymphatic channels by similar cells in nearly all cases, and occurrence of renal angiomyolipoma in 30 to 50% of S-LAM cases (3). Past seminal publications identified mutations and loss of heterozygosity (LOH) in TSC2 in pulmonary LAM cells of four S-LAM patients who had angiomyolipomas with known mutations (6), and in two of five S-LAM patients (7). Other studies have shown that TSC2 mutations are seen in cultured LAM cells (8), and that TSC2 LOH is seen in LAM cells isolated from blood, urine, chylous fluid, and bronchoalveolar lavage fluid in 50 to 90% of S-LAM patients (9, 10). Some of the results of this study have been previously reported in the form of abstracts (11, 12).

We searched for TSC2 mutations in 10 S-LAM patients that had lung transplant using a combination of laser capture microdissection (LCM) and next-generation sequencing (NGS) (see online supplement). Examination of formalin-fixed paraffin-embedded sections showed classic features of LAM, with multiple cysts and disseminated LAM nodules, the cells of which were smooth muscle actin (SMA) and HMB45 positive (Figures E8A–E8G in the online supplement). LCM guided by SMA immunohistochemistry was performed on frozen sections to collect LAM cells from nodules (Figure E1) and avoid inclusion of lymphatic channels, lymphocytes, and other cell types (Figures E3 and E4). After DNA extraction and amplification, NGS was performed using either the 454 or Illumina platforms (Reference 13 and online supplement). High read depth was achieved across the coding region of TSC2 (median, 486; >200 in 96% of exons), enabling detection of low-frequency sequence variants. Nine different pathogenic sequence variants were detected in TSC2 in eight different sporadic LAM samples, at frequencies ranging from 4 to 60% (Table 1), with most seen at a frequency less than 20%. All were verified by Sanger or SNaPshot analysis on unamplified DNA (Figures E6 and E7). Four samples showed evidence of two-hit inactivation in TSC2, three of them with a mutation and LOH, and one with two point mutations. The four cases without LOH all had low allele fractions for mutation in TSC2 (≤16%), making detection of LOH difficult. The two LAM samples in which no TSC2 mutation was identified also showed no mutation in TSC1 by NGS sequencing. Significantly, in one of these cases, over 90% of LAM cells were positive for HMB45 (case 9023) (Figure E8B), and in the other, the patient had a concomitant renal angiomyolipoma, providing additional evidence of S-LAM (1, 2). HMB45⁺ cells in LAM sections have relatively low PCNA expression, suggesting that they are less proliferative than HMB45⁻ cells (14). Although some types of inactivating mutations are missed by exonic sequencing, immunohistochemistry for TSC2 (tuberin) and TSC1 (hamartin) on the two cases without defined mutations showed strong expression of each protein, whereas little or no expression of either was seen in a control case with two mutations in TSC2 (Figure 1 and Figure E10). Furthermore, the two cases without TSC1/TSC2 mutation showed no expression of phospho-S6 kinase by immunohistochemistry, in contrast to cases in which biallelic TSC2 mutations were identified (Figures 1G–1I). This suggests that mechanistic target of rapamycin complex 1 (mTORC1) was not activated in the LAM nodules of these two patients, consistent with normal TSC1/TSC2 function and mTOR regulation.

TABLE 1.

MUTATIONS AND SINGLE-NUCLEOTIDE POLYMORPHISMS DETECTED IN TSC1 AND TSC2 IN 10 SPORADIC LAM SAMPLES

Sample	Method	TSC2 Mutation	Allelic Frequency	Mutation Effect	TSC2 SNPs	LOH	MLPA Copy #	TSC1	HMB45 Expression*
9016	Illumina	c.5024C > T	16%	p.1675P > L missense	2	No	ND	ND	5+
9020	454	c.781C > T	50%	p.261R > W missense	1	No	NL	ND	1+
	454	c.3610+1G > A	15%	Splice
9022	454	c.789_806del18	8%	In-frame deletion	0	No	NL	ND	3+
9023	454	None			3	No	NL	None†	5+
9030	454	None			1	No	NL	None	1+
9034	454	c.5127delC	39%	Truncation	4	No	NL	ND	1+
					2580T>C	Yes‡
9036	454	c.1947-4_2030del88	60%	Truncation	1	Yes§	NL	ND	1+
9038	454	c.1837C>T	4%	p.Q613X	1	No	NL	ND	2+
9043	Illumina	c.3167delG	18%	Truncation	14	Yes¶	NL	ND	1+
9059	Illumina	c.1513C>T	16%	p.R505X	5	No	NL	ND	1+

Definition of abbreviations: LAM = lymphangioleiomyomatosis; LOH = loss of heterozygosity; MLPA = multiplex ligation-dependent probe assay; ND = not done; NL = normal; SNPs = single-nucleotide polymorphisms.

^*HMB45 expression is scored on a qualitative scale from 1+ (weak) to 5+ (strong).

^†This sample had one TSC1 SNP.

^‡This sample showed a 2:1 allelic ratio in sequencing reactions, consistent with LOH, but only for one of five heterozygous SNPs found in this sample.

^§This sample showed evidence for LOH by Sanger sequencing of the deletion mutation, in that there was increased frequency of the deletion allele in comparison to the wild-type allele.

^¶LOH was seen in this sample, with a skewed allele ratio for 14 TSC2 intragenic SNPs, for which the average minor allele frequency was 0.42 in comparison to an average minor allele frequency of 0.47 for control sample SNPs (P < 0.001).

Figure 1.

Hamartin and tuberin expression and lack of phosphorylated S6K (p-S6K) expression in lymphangioleiomyomatosis (LAM) nodules from two LAM cases with no mutation identified. Three LAM lesions are shown. (A–D) Two cases, 9023 and 9030, had no mutations found in either TSC1 or TSC2, and expressed hamartin and tuberin. (E and F) In contrast, case 9036 had no expression of those proteins, and had a high level of TSC2 mutation with LOH. (G–I) Cases 9023 (G) and 9030 (H) do not express p-S6K, whereas case 9036 (I) shows strong p-S6K immunopositivity. (J and K) Hamartin and tuberin expression in uninvolved lung parenchyma and (L and M) in prostate carcinoma (from tissue array control included in every slide). (N–P) p-S6K expression in uninvolved lung parenchyma (note + reactive alveolar epithelium in upper left corner), prostate carcinoma (negative), and breast carcinoma (positive), respectively. Original magnifications, ×100.

The occurrence of TSC2 mutations at relatively low frequency, or not at all, in these LAM samples is surprising. As shown here and elsewhere (13), NGS has the capability to detect mutations that are present at a very low allelic fraction, 1% or lower, enabling high sensitivity analyses as performed here. High sample contamination with non-LAM cells seems unlikely given our efforts at histological purification of the cells collected by LCM, the identification of mutations at high frequency in some cases, and the detection of a low- and a high-frequency mutation in one case. It is possible that TSC2 mutations occur in a subset of LAM cells, after other unknown initiating events, and are not the primary driver of LAM development, or TSC2-mutant cells recruit stromal cells to adopt an SMA-positive phenotype. In addition, the two cases without TSC1 or TSC2 mutations identified also suggest that alternative genetic mechanisms may be operative in some cases of LAM.

Recently the Multicenter International LAM Efficacy of Sirolimus (MILES) trial demonstrated that rapamycin (sirolimus), an mTORC1 inhibitor that blocks the pathway activated by loss of tuberin (15), was an effective therapy for treatment of LAM (16). However, it is notable that continuing FEV₁ decline was seen in 50% of patients on rapamycin therapy. It is possible that some of these nonresponders do not have loss of TSC2 with activation of mTORC1 as a fundamental pathogenic mechanism for LAM development, consistent with our data.