Introduction

Anaplastic lymphoma kinase (ALK) is a well-known driver oncogene in anaplastic large cell lymphoma (ALCL), non-small cell lung cancer (NSCLC) and renal cell carcinoma (RCC) and has been identified in other types of tumors, including inflammatory myofibroblast tumor (IMT), epithelioid inflammatory myofibroblast sarcoma (EIMS), and epithelioid fibrous histiocytoma (EFH). IMT is a rare neoplasm characterized by spindle-cell proliferation within an inflammatory infiltrate [1]. Approximately 50 to 60% of IMTs have activating rearrangements of the ALK gene on chromosome 2p23, resulting in constitutive tyrosine kinase activation, and crizotinib has proven to be an effective therapeutic approach. Recently, immunohistochemical expression of ALK in the majority of angiomatoid fibrous histiocytomas (AFHs) was demonstrated in an initial series by Cheah et al. and subsequently confirmed by Van Zwam et al. [2, 3]. Interestingly, the expression of ALK did not seem to be associated with ALK rearrangement or copy number variations. However, there are very few reports regarding the aberrant expression of ALK in gastrointestinal stromal tumors (GISTs).

GIST is the most common type of mesenchymal tumor and arises from Cajal cells in the gastrointestinal tract, with an annual incidence rate of 10–20 per million [4, 5]. Indeed, with the identification of activating mutations of either c-KIT or platelet-derived growth factor receptor (PDGFRA) tyrosine kinases as main players in GIST pathogenesis and development, in a few years, imatinib has become the backbone for the treatment of unresectable and advanced GISTs, the efficacy of which is profoundly affected by the underlying tumor genotype [6]. Tumors with the PDGFRA D842V mutation, which accounts for the majority of PDGFRA-mutant GISTs, are primary imatinib resistant but have been reported to be sensitive to avapritinib [7,8,9,10]. Recently, our group reported a case of PDGFRA-mutant GIST with ALK expression. According to the clinicopathological features and the National Institutes of Health (NIH) classification, the risk of recurrence was classified as high [11]. Herein, we aimed to investigate the association between the expression of ALK and the clinicopathological characteristics of GISTs.

Materials and methods

Patients and tumor samples

We enrolled a total of 506 patients who had a confirmatory diagnosis of GIST after surgical resection or biopsy and were all sequenced by the Sanger sequencing method in the Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, from January 2015 to June 2022. The diagnoses were all confirmed through histopathological, immunohistochemical, and molecular tests. Hematoxylin and eosin (HE)-stained slides and formalin-fixed paraffin-embedded (FFPE) tissue blocks were obtained from the Department of Pathology. The study was approved by the Ethics Committee of Union Hospital (Ethical Approval Number: S-377). All patients signed informed consent forms before the initiation of the study-related procedure.

Tissue microarray

The tissue microarray (TMA) technique was performed as described previously [12, 13]. HE slides were screened to identify optimal tumor tissue for analysis. A red pen was used to mark representative areas on the ‘donor’ FFPE GIST blocks. TMA blocks were then constructed (DAN JIER Electronics Co., Ltd, JiNan, China). Three core tissue biopsies (diameter, 1.0 mm) were collected from non-necrotic areas of tumor foci in each individual ‘donor’ FFPE GIST block and precisely arrayed into a new ‘recipient’ paraffin-embedded block (24 × 37 mm) using a custom-built instrument. The TMA was placed on top of a glass slide and heated to 55 °C for 15 min to securely bind and level the cores to the ‘recipient’ block. The glass slide and TMA block ensemble were then transferred to an ice pack for 30 min to complete construction. Sections (4 μm) of the ‘recipient’ TMA blocks were made using standard techniques. Due to the biopsy GIST samples lacking sufficient tumor tissues, the TMA technique was performed only in surgical GIST samples. The biopsy GIST FFPE tissue blocks were cut into 4 μm sections for immunohistochemistry staining.

Immunohistochemistry

Paraffin Sects. (4 μm) for immunohistochemistry were cut from ‘recipient’ TMA blocks and biopsy GIST FFPE tissue blocks. Immunohistochemical staining was performed by the Envision two-step method. The expression of ALK tyrosine kinase was examined using two monoclonal antibodies against ALK (Ventana Medical System, Inc.; Clone: 1A4 and D5F3). The other immunohistochemical indicators referred to archived sections, including CD117 (Ventana Medical System, Inc., Tucson, AZ, USA; Clone: c-kit), DOG1 (MXB, Fuzhou, China; Clone: SP31), CD34 (Gene Tech, Shanghai, China; Clone: QBEnd 10), desmin, smooth muscle actin (SMA) (Gene Tech; Clone: 1A4), SDHB (ZSGB-BIO, Beijing, China; Clone: OTI1H6), S-100 protein (MXB, Fuzhou, China; Clone: 4C4.9), and Ki-67 (MXB, Clone: MIB-5).

Fluorescence in‐situ hybridization

Fluorescence in situ hybridization (FISH) was performed on 4‐μm‐thick FFPE sections obtained from ALK IHC-positive patients using the Vysis dual-color ALK break-apart probe kit (Abbott Molecular/Vysis, Des Plaines, IL, USA) following the manufacturer’s instructions. The ALK gene copy number was counted in a minimum of 50 nonoverlapping nuclei in select cases.

Sanger sequencing

The Sanger sequencing method was used to detect c-KIT (exon 9, 11, 13, and 17) and PDGFRA (exon 12 and 18) gene mutations in all cases. According to the kit (c-KIT and PDGFRA gene mutation detection kit of Xiamen Aide Co., Ltd, Xiamen, China) instructions, genomic DNA was extracted from the samples. C-KIT exons 9, 11, 13, and 17 and PDGFRA exons 12 and 18 were amplified with polymerase chain reaction (PCR). After detecting the PCR products by gel electrophoresis, the PCR products were sequenced using ABI 3730XL DNA sequencers. The sequencing results were compared with the standard template sequences of the BLAST program within the CHROMAS software to identify the gene mutation loci.

Next-generation sequencing

FFPE tissue blocks for next-generation sequencing (NGS) were obtained from ALK IHC-positive patients. NGS was performed by Tongshu Gene (Shanghai, China). The OncoPanel consists of targeted sequencing of 556 cancer-related genes. The 23 fusion genes in the 556 panel were detected by DNA sequencing, including ALK rearrangement, amplifications or activation mutations.

Reverse transcription-polymerase chain reaction

For comparison between the tumor and normal tissues, ALK expression in tumor tissues was calculated using normal tissue sample expression as a reference control. Only in cases 1 and 3, tumor tissues and normal control tissues were available. FFPE tissue blocks for reverse transcription-polymerase chain reaction (RT–PCR) were obtained from surgically resected specimens in cases 1 and 3. For each case, tumor tissues and adjacent normal tissues were randomly selected in this experiment. Total RNA was extracted from paraffin-embedded tissue sections and isolated by an RNA extraction kit (AmoyDx, Xiamen, China) according to the manufacturer's instructions. Then, cDNA was obtained via reverse transcription (PrimeScriptTM II 1st Strand cDNA Synthesis Kit, TaKaRa, Dalian, China). PCR was performed using 2 × Hieff ® Robust PCR Master Mix (With Dye) (Yeasen, Shanghai, China). The primers included human ALK, forward 5′- AACCACTTCATCCACCGAGACATTG -3′ and reverse 5′- TGTTGCCAGCACTGAGTCATTATCC-3′. Human glyceraldehyde-3-phosphate dehydrogenase (GAPDH) was used as an endogenous reference gene primer to normalize human ALK expression. The expected lengths of the PCR fragments were 377 and 138 bp. The cDNA (5 µl/lane) was separated by 40% polyacrylamide gel electrophoresis (PAGE). Five microliters of DNA marker liquid (code no. 3590Q; TaKaRa, Dalian, China) was used to separate PCR products, the results of which were observed using a gel imaging analyzer (Tanon, Shanghai, China). Due to the lack of adjacent normal tissues as a reference, the patients in cases 2 and 4 did not undergo RT–PCR.

Follow-up and statistical analysis

The 52 PDGFRA-mutant patients were followed up by telephone and the outpatient service at regular intervals (3- to 6-month intervals postoperatively for 2 years and annually thereafter). Overall survival (OS) was defined as the time from diagnosis to death. Statistical analysis was conducted in SPSS 26.0 for Windows (SPSS, Chicago, IL, USA). The median or mean ± SD was used to express the measurement data, and overall survival curves were drawn by the Kaplan–Meier (K-M) method.

Results

Types of gene mutations

Figure 1 summarizes our study design. A total of 506 GIST patients were enrolled in our study. After sequenced by Sanger sequencing method, the mutation types were identified: 426 patients with c-KIT mutation, 52 patients with PDGFRA mutation and 28 patients with wild-type mutation. Employed with TMA and IHC, four ALK-positive cases and 502 ALK-negative cases were revealed by TMA and IHC. To explore the molecular basis of ALK expression, FISH and NGS were performed in four ALK-positive cases. Among them, two were conducted with RT‒PCR.

Fig. 1
figure 1

Flow diagram of the study design

Full size image

Among the 506 GIST patients, the c-KIT mutation was predominant (426, 84.2%), followed by the PDGFRA mutation (52, 10.3%), while the wild-type was the least (28, 5.5%). Deletion and point mutation were the most frequent mutational patterns (30.2%, 153/506 for both), followed by mixed (16.2%, 82/506) and duplication (13.4%, 68/506). Insertion was relatively fewer, accounting for 2.0% (10/506). In the c-KIT mutation, deletion in exon 11 occurred most frequently (29.6%, 150/506), while in the PDGFRA mutation, the most frequent type was mutations in exon 18 (9.1%, 46/506), followed by mutations in exon 12 (1.2%, 6/506). The detailed mutation types of all patients are summarized in Supplementary Table S1.

ALK protein expression and gene variant status

ALK (1A4 and D5F3) immunohistochemical stains were used in 506 GIST cases, of which 4 cases showed strong positivity. The positive cases were all identified in PDGFRA mutation (7.7%, 4/52), while none in the c-KIT mutation or wild-type mutation. No ALK rearrangement, amplifications or activation mutations were identified for the 4 ALK IHC-positive cases according to the DNA-based sequencing, but one was identified as amplification by FISH.

Clinicopathological features of 4 ALK-positive patients

The 4 ALK IHC-positive patients were all male. The patients’ median age was 55.5 years old (range from 37 to 66 years old). Among them, 3 (75%, 3/4) patients were younger than 60 years old. The tumors of 4 ALK-positive patients all occurred outside the stomach. The maximum sizes of the tumors were all greater than 10 cm, and the mitotic accounts of 3 ALK-positive patients (75%, 3/4) were more than 10/5 mm2. The predominant patterns of growth were epithelioid (2/4), spindle and mix type (1/4). They were all identified as high-risk classification according to NIH criteria, 2008 modified version. A review of the clinicopathological features and prognosis of 4 ALK-positive patients is provided in Table 1. And the demographic and clinicopathological features of ALK-positive group and ALK negative group were summarized in Supplementary Table S2.

Table 1 The clinicopathological features and prognosis of 4 ALK-positive patients
Full size table

CT imaging, morphology, IHC, and FISH of the 4 ALK-positive patients

The examination images of case 1 are shown in Fig. 2. Computed tomography (CT) revealed an uneven density shadow sized 16.2 × 15.4 × 8.8 cm on the right upper abdominal cavity (Fig. 2A). Postoperative pathological examination revealed a large number of epithelioid tumor cells (Fig. 2B) with regional cystic degeneration, hemorrhage and necrosis. Immunohistochemical staining showed positive expression of CD117 (weakly) and DOG-1 and negative expression of SMA (Fig. 2C, D, E). ALK expression was strong in tumor cells with two different antibody clones, 1A4 (Fig. 2F) and D5F3 (Fig. 2G). Fluorescent in situ hybridization (FISH) using a break-apart probe showed no ALK rearrangement (Fig. 2H). Information on cases 2–4 was shown in the corresponding supplementary material (Supplementary Figures S1–S3).

Fig. 2
figure 2

The examination images of case 1. A Abdominal CT scan shows an uneven density shadow sized 16.2 × 15.4 × 8.8 cm on the right upper abdominal cavity (red arrow). B The tumor was consisted of epithelioid tumor cells by H&E staining (200 ×). C–E (200 ×) The tumor cells showed a positive signal for CD117 (C) and DOG1 (D) and a negative signal for SMA (E). F–G ALK IHC showed strong positive staining for 1A4 (F) and D5F3 (G). K FISH did not reveal an ALK rearrangement (1000 ×)

Full size image

Relative ALK mRNA expression

The electrophoresis results of case 1 and case 3 were shown in Supplementary Figure S4. Electrophoresis analysis showed uniform density between the band in lane 3 and the band in lane 4, which lay in the < 150-bp area and were consistent with the theoretical amplification length (138 bp). In Lane 1, no band was observed. Compared with lane 1, a faint band at ≈ 400 bp (theoretical amplification length 377 bp) was observed in lane 2. The bands in lanes 7 and 8 also exhibited a uniform density, located at ≈ 150 bp. In lane 5, no association of the ALK band was detected. In lane 6, a faint band appeared at ≈ 400 bp (theoretical amplification length 377 bp).

Survival of 52 PDGFRA-mutant patients

The 52 patients were followed until death or loss to follow-up (F/U). The median OS of the ALK-positive group (n = 4) and ALK-negative group (n = 48) was 25.00 months and not reached, respectively. Three ALK-positive patients (75%) died at 2, 25, and 49 months after diagnosis, and 2 ALK-negative patients (3.6%) died at 3 and 6 months after diagnosis. The OS was shown in K-M curves (Fig. 3).

Fig. 3
figure 3

The K-M plot for overall survival of ALK-positive group and ALK-negative group

Full size image

Discussion

ALK, a transmembrane tyrosine kinase, belongs to the superfamily of insulin receptors that regulate cellular growth and may trigger neoplastic transformation, including platelet-derived growth factor receptors, epidermal growth factor receptor, human epidermal growth factor receptor type 2, and insulin-like growth factor-1 receptors [14]. Aberrantly formed ALK oncogenes, mainly caused by fusion mutations, ALK gain-of-function mutations, or ALK amplification, have been identified in various cancers. However, the aberrant expression of ALK in GIST has rarely been reported. Recently, our group reported a case of PDGFRA-mutant GIST with ALK expression, in which the tumor had an immense size, and follow-up revealed liver metastases that developed quickly [11]. In this study, a total of 506 GIST patients were included. ALK-positive expression was not identified in c-KIT-mutant or wild-type GISTs, while the rate of ALK expression among PDGFRA-mutant GISTs was 7.7% (4/52) without aberrant ALK mutations. Furthermore, the ALK-positive expression cases exhibited more aggressive biological behavior and a worse overall prognosis.

PDGFRA-mutant GISTs preferentially occur in the stomach, occasionally outside the stomach, including the intestine, abdominal cavity, and mesentery [15]. The histological types of PDGFRA-mutant GISTs are mainly epithelioid or mixed subtype and are rarely spindle cell type [15]. Immunostaining often shows poor expression of CD117 and DOG1 [16]. ALK expression in GISTs with PDGFRA mutations can potentially lead to difficulty in differentiating them from IMTs, more than 50% of which displayed ALK gene rearrangement and overexpression. Here, we reported that 7.7% of GISTs with PDGFRA mutations showed strong and diffuse staining for ALK, which was confirmed by antibodies from two different clones in our research. Therefore, GISTs should also be taken into account in the differential diagnosis of ALK-positive soft tissue tumors, especially IMT, because the suggested treatment regimens for IMT and GIST are strikingly different. Therefore, genetic testing is crucial for differential diagnosis.

In general, studies focused on ALK gene alterations and expression in GISTs are rare. Zhao Lei et al. reported a GIST occurred in the small bowel with the ALK protein overexpression. The NGS sequencing identified a rearrangement involving PPP1R21 and ALK. No relapse was observed at the 8-month follow-up for the patient [17]. Recently, Huang et al. [18] reported a quadruple wild-type (qWT) GIST with ALK (D5F3) overexpression, and a novel CDC42BPB-ALK fusion was identified by NGS. However, the two cases both lacked known GIST driver genes, such as KIT, PDGFRA, BRAF, and the SDHx gene family. Thus, we think it is difficult to rule out the possibility of ALK-rearranged mesenchymal tumors with the Cajal cell phenotype and positive expression of CD117 and DOG-1.

To the best of our knowledge, there are few reports about ALK protein expression without ALK gene abnormalities in soft tissue tumors. Similarly, John R. Goldblum and colleagues reported a high frequency (9/11) of ALK positivity in AFH [17], which is an EWSR1 (or rarely FUS) rearrangement-associated neoplasm characterized by accompanying lymphoplasmacytic infiltrate. However, the molecular basis for ALK immunoreactivity in AFH in their series is unclear. In our study, two available cases of pairwise comparison RT–PCR revealed increased ALK mRNA expression in tumors compared with adjacent normal tissues but without ALK gene rearrangement, amplifications or activation mutations detected by DNA-based NGS. Therefore, we speculated that PDGFRA mutation was probably involved in upregulating the expression of the ALK gene, and the underlying molecular mechanisms remain to be explored.

GISTs with PDGFRA mutations are believed to involve an indolent process and exhibit a favorable prognosis [19, 20]. However, our team recently reported the first case of ALK expression in a PDGFRA-mutated GIST, and follow-up revealed liver metastases that developed quickly [11]. According to our limited data, GIST patients with both ALK expression and PDGFRA mutation tend to have a larger tumor diameter, higher recurrence risk and more mitotic figures, indicating that ALK expression may be associated with a worse prognosis. Our follow-up results were broadly consistent with the above speculation. Three of 4 ALK-positive patients presented progression within the first 2 years, and one of these patients with PDGFRA D842V mutation experienced liver metastasis only 6 months after surgical treatment. In general, ALK-positive patients exhibited invasive biological behavior and a worse prognosis. The patient in case 1 responded poorly to the second-generation ALK-TKI ceretinib after treatment with first-line dasatinib. We presumed that ALK expression might be caused by epigenetic modification or other mechanisms rather than the main driver gene, which results in disease progression.

Indeed, our study has several limitations, including the retrospective, single-center design. To our knowledge, it was probably inappropriate to statistically compare the only 4 patients with ALK-positive disease with the remaining PDGFRA mutant cases, which may result in some error in the reliability and accuracy of the statistical results for survival analysis. In addition, the mechanism of ALK expression and the unfavorable outcome in PDGFRA mutation patients is unclear.

In summary, our study explored the ALK expression status in GISTs and revealed ALK overexpression in 7.7% PDGFRA-mutant GISTs specifically. The results indicated that PDGFRA-mutant GISTs should be considered a differential diagnosis when encountering an ALK-positive mesenchymal tumor, especially those with CD117-negative or weakly positive in immunohistochemical staining. If necessary, molecular detection should be carried out for further clarification. In addition, the expression of ALK may be a factor leading to more invasive biological behavior and poor prognosis in ALK-positive PDGFRA-mutated GISTs.