Introduction
Globally, lung cancer is the primary contributor to cancer-associated mortality and the leading cause of mortality in the majority of regions (1-4). An estimated 222,500 novel lung cancer cases and 155,870 mortalities were predicted to have occurred in 2017 in the USA (1). Of all patients with lung cancer, those with non-small cell lung cancer (NSCLC) account for ~80% (5). The identification and investigation of genetic drivers such as epidermal growth factor receptor (EGFR)-activating mutations, have contributed to a gradual decrease in lung cancer-associated mortality (3,6,7). EGFR is a member of a larger family of transmembrane receptor tyrosine kinases (TKs) that activate cell proliferation and survival (8). Mutations in the TK domain of EGFR in NSCLC exhibit improved responses to EGFR tyrosine kinase inhibitors (TKIs) such as gefitinib and erlotinib (7), particularly exon 19 deletions and L858R in exon 21.
Previous studies have identified that female Asian patients without a history of smoking and with adenocarcinoma histology are more likely to exhibit EGFR mutations (9). Therefore, validating the EGFR genotype status in patients with NSCLC may help to select those who will benefit from TKIs when making treatment decisions. However, inaccessible tumor sites, insufficient tissues for testing, heterogeneous tumors and a patient’s refusal to undergo invasive detection all pose limitations to performing the individual genotype test. Thus, developing non-invasive and effective methods to help with identification of the status of the EGFR gene is required.
[18F]fluorodeoxyglucose (FDG) positron emission tomography (PET)-computed tomography (CT), which is based on high glucose metabolism in lesions, serves an important function in initial staging, evaluating the response following therapy and radiation therapy planning during the management of NSCLC (10,11). Therefore, as a non-invasive method, the quantification of glucose metabolism using FDG-PET is one way to predict EGFR mutations. The standard uptake value maximum (SUVmax), a metabolic parameter from PET for FDG uptake, is associated with prognosis in NSCLC and previous studies revealed that patients with NSCLC with a low SUVmax for the primary lesion tend to have better outcomes (12,13), indicating that a low SUVmax may be associated with EGFR gene mutations. However, in clinical practice, studies that aim to reveal the FDG uptake and EGFR mutation status are controversial, and the potential mechanisms by which EGFR mutations alter FDG uptake remain largely unknown; therefore, further clinical studies and investigations of the underlying molecular mechanisms should be performed.
Glucose transporter 1 (GLUT1) serves crucial functions in FDG uptake (14,15); furthermore, GLUT1 expression can be altered by dysregulated reactive oxygen species (ROS) activity (16,17), in which NADPH oxidase 4 (NOX4) is primarily responsible for ROS production (18). Considering the aforementioned studies, we hypothesized that EGFR mutations may regulate FDG uptake via the NOX4/ROS/GLUT1 axis in NSCLC.
In the present study, the association between EGFR mutations and SUVmax was investigated, the receiver operating characteristic (ROC) curve was analyzed to identify the optimum cut-off value for SUVmax in predicting EGFR mutation, GLUT1 expression and ROS activity were determined in the A549 and PC-9 (EGFR mutation, 19del) cell lines, and NOX4 mRNA and protein expression were investigated to test the hypothesis. Subjects were recruited and enrolled in the present study, and subjected to a battery of tests that included FDG-PET-CT scanning and EGFR mutation testing. The study flow chart is presented in Fig. 1.
Materials and methods
Patients and diagnosis
In total, 157 patients (median age 65.8 years; range, 48-81 years) with NSCLC who were diagnosed at the Department of Pathology of The Third Affiliated Hospital of Kunming Medical University (Kunming, China) from June 2015 to October 2017 were enrolled in the present study. All patients fulfilled the following entry criteria: i) The diagnosis was made histologically, and the patients underwent EGFR gene testing; ii) PET-CT was performed prior to any therapy; iii) complete clinical information was obtained; iv) histopathology was reviewed at Yunnan Cancer Hospital (Yunnan, China); and v) written informed consent was obtained from the patients. The study protocol was approved by the Ethics Committee of The Third Affiliated Hospital of Kunming Medical University. All procedures performed in the present study that involved human participants were with the approval of the Institutional Review Board of The Third Affiliated Hospital of Kunming Medical University and in accordance with the 1964 Declaration of Helsinki and its later amendments or comparable ethical standards.
EGFR mutation analysis
An AmoyDx® EGFR 29 Mutations Detection kit (Amoy Diagnostics Co., Ltd., Xiamen, China) was used to detect the EGFR mutation in DNA extracted from tissue and plasma samples using a Qiagen DNA mini-kit (Qiagen GmbH, Hilden, Germany), according to the manufacturer’s protocol. The kit methodology is based on amplification refractory mutation system (ARMS) technology, which was used to detect 29 mutations in exons 18 to 21 of the EGFR gene (19) All ARMS primer pairs (AmoyDx, Super-ARMS, 19 del, forward, 5ʹ-GTTAAAATTCCCGTCGCTATCAAGACATCT-3ʹ, and reverse, 5ʹ-CACAGCAAAGCAGAAACT CACAT-3ʹ; L858R, forward, 5ʹ-GCAGCATGTCAAGATCACAGATTTTGGGCG-3ʹ, and reverse, 5ʹ-GTCAGGAAAATGCTGGCTGACCTAAAG-3ʹ; T790M, forward, 5ʹ-CTCACCTCCACCGTGCARCTCATCAT-3ʹ, and reverse, 5ʹ-CAATATTGTCTTTGTGTTCCCGGACA-3ʹ; G719X, forward, 5ʹ-CTCACCTCCACCGTGCARCTCATCAT-3ʹ, and reverse, 5ʹ-CCGTGCCGAACGCACCGGAGCA-3ʹ; S790I, forward, 5ʹ-AGCGTGGACAACCCCCACCAC-3ʹ, and reverse, 5ʹ-CCGTGCCGAACGCACCGGAGCA-3ʹ) were used for polymerase chain reaction (PCR), with the following criteria: Concentration of 1 mmol/l, control reaction primers at a concentration of 0.1 mmol/l. PCR was performed with denaturation at 94°C for 30 sec, 30 cycles of 95°C for 30 sec, 55°C for 30 sec and 72°C for 30 sec, and another 72°C for 6 min.
Interpretation and image analysis of FDG-PET-CT scans
FDG-PET-CT scan images were acquired in the department of PET-CT Center of Yunnan Cancer Hospital using the syngo. via platform (Siemens Healthineers, Erlangen, Germany) (slice thickness, 3-5 mm). The patients fasted for a minimum of 6 h, an FDG dose of 12 mCi was administered, and the patients were scanned from the skull base to the mid-thigh using multiple bed positions (two or three bed positions; acquisition time, 2 min/bed position) 1 h after injection. CT-attenuated data were reconstructed using ordered subset expectation maximization for the two scanner sites. Representative images are presented in Fig. 2A and B. The images were reviewed by two board-certified nuclear medicine physicians with 2 and 10 years of experience, respectively. A syngo MultiModality WorkPlace system (Siemens Healthineers) was used to select and measure structures throughout the body using the region-of-interest (ROI) tool within the software. Circular ROIs with a diameter of 10 mm were drawn on transaxial FDG-PET-CT images using the fusion CT scan as an anatomical guide.
Cell culture
Human NSCLC A549 and NCI-H1975 cells were purchased from the American Type Culture Center (Manassas, VA, USA). PC-9 cells were purchased from RIKEN Cell Bank (Tsukuba, Japan) and is a 19del-positive cell line, whereas A549 is a cell line expressing wild-type EGFR, and the NCI-H1975 cell line harbors the L858R and T790M substitution EGFR mutations. NCI-H1975 and PC-9 cells were grown in RPMI-1640 medium (Thermo Fisher Scientific, Inc., Waltham, MA, USA) supplemented with 10% fetal bovine serum (FBS; Hyclone; GE Healthcare, Logan, UT, USA), 2 mM L-glutamine and 1% penicillin/streptomycin. A549 cells were cultured in Dulbecco’s modified Eagle’s medium (Thermo Fisher Scientific, Inc.) supplemented with 10% FBS, 2 mM L-glutamine and 1% penicillin/streptomycin. All cells were maintained and propagated as monolayer cultures at 37°C in a humidified 5% CO2 incubator.
NOX4 mRNA determination
Total RNA was extracted from A549 and PC-9 cells using the TRIzol® reagent (Thermo Fisher Scientific, Inc.), and was reverse-transcribed using a SuperScript II Reverse Transcriptase kit (Takara Biotechnology Co., Ltd., Dalian, China), according to the manufacturer’s protocol. Quantitative PCR (qPCR) was performed using a SYBR Green Supermix kit (Takara Biotechnology Co., Ltd.) and the ABI 7300 detection system (Thermo Fisher Scientific, Inc.). Blank controls with no cDNA templates were included to rule out contamination. The specificity of the PCR product was confirmed by melting curve analysis and gel electrophoresis. All gene expression levels were normalized to that of the housekeeping gene U6. Relative expression levels of the target gene normalized to U6 were calculated using the 2‐ΔΔCq method (20). Each reaction was performed independently at least three times. The following primer pairs were used: NOX4 primer set, 5ʹ-TGTTGGGCCTAGGATTGTGTT-3ʹ (forward) and 5ʹ-AGGGACCTTCTGTGATCCTCG-3ʹ (reverse); U6 primer set, 5ʹ-CTCGCTTCGGCAGCACA-3ʹ (forward) and 5ʹ-AACGCTTCACGAATTTGCGT-3ʹ (reverse). PCR was performed using the following parameters: 95°C for 5 min, 30 cycles of 94°C for 30 sec, 58°C for 30 sec and 72°C for 30 sec, and 72°C for 5 min.
Western blot analysis
The cells were solubilized in ice-cold radioimmunoprecipitation assay lysis buffer. Amounts of 25 µg protein (determined using a Bicinchoninic Acid Protein assay kit from Abcam, Cambridge, UK) from the cytosolic fraction were separated by SDS-PAGE (10% gel) and transferred onto a polyvinylidene difluoride membrane. The membrane was incubated with 5% skimmed milk in Tris-buffered saline containing 0.2% Tween-20 at 37°C for 2 h. The membrane was then incubated with rabbit anti-NOX4 (cat. no. ab79971; 1:1,000 dilution) and mouse anti-β-actin (cat. no. ab8226; 1:2,000 dilution) primary antibodies (both from Abcam) at room temperature for 2 h. Following washing four times with PBS containing 0.2% Tween-20 (PBST) each for 10 min, the membrane was incubated with goat anti-rabbit (cat. no. sc-2030; 1:1,500 dilution) and goat anti-mouse (cat. no. sc-2005; 1:2,000 dilution) secondary antibodies (Santa Cruz Biotechnology, Inc., Dallas, TX, USA) at 4°C overnight. Following washing four times with PBST each for 10 min, proteins recognized by the antibody were visualized with the Luminata Forte Western Blotting Substrate (EMD Millipore, Billerica, MA, USA), according to the manufacturer’s protocol. Image-Pro Plus software (version 6.0) was used to analyze the relative protein expression, represented as the density ratio against β-actin, which was used as an internal reference.
ROS detection
Intracellular ROS levels were determined using the oxidative-sensitive fluorescent probe dihydroethidium (DHE; Molecular Probes; Thermo Fisher Scientific, Inc.), as described previously (21) with certain modifications. Briefly, A549, PC-9 and NCI-H1975 cells (2.5×105) in 6-well plates were incubated with 4 M DHE at 37°C for 45 min. The cells were harvested and washed with PBS. The fluorescence from oxidized DHE was detected at a wavelength of 630 nm and fluorescence images were captured using an Olympus BX51 fluorescence microscope (Olympus Corporation, Tokyo, Japan).
Statistical analysis
Categorical covariates were analyzed using Pearson’s χ2 test or Fisher’s exact test as appropriate, and continuous covariates were analyzed using Student’s t-test or analysis of variance, as appropriate. A ROC curve was generated to determine a cut-off for the SUVmax of the primary tumor. Multivariate logistic regression analysis was performed to test the variables that yielded predictors of EGFR mutations. The area under the curve (AUC) was used for the predictive value. P<0.05 was considered to indicate a statistically significant difference. GraphPad Prism (version 6.0; GraphPad Software, Inc., La Jolla, CA, USA) was used for the analysis.
Results
Clinical features and EGFR mutations
The baseline characteristics of the patients are listed in Table I. There were 157 patients (84 males and 73 females) that met the eligibility criteria. Of those, 54 patients (34.3%) were EGFR mutation-positive. Exon 19 deletion and L858R in exon 21 were the most common mutations, accounting for 48% (26 patients, including 3 combined mutation types) and 33.3% (18 patients, all single mutation), respectively. Other mutation types were single G719X (3 patients, 5.5%), single T790M (5 patients, 9.2%), single S768I (2 patients, 3.7%) and combined 19del+T790M (3 patients, 5.5%). The EGFR mutations were more frequent in female patients compared with in male patients (42.3 vs. 26.6%; P=0.045). The median age was 58.3 years, and 96 patients (61.1%) had a history of smoking. EGFR mutations were more frequent in non-smokers compared with in smokers (49.1 vs. 19.1%; P=0.006). There were 144 patients (91.7%) with adenocarcinoma and the remaining 13 patients were without adenocarcinoma (8.3%). EGFR mutation status was more frequent in patients with adenocarcinoma compared with patients without adenocarcinoma (36.8 vs. 7.7%; P=0.036). In addition, patients harboring EGFR mutations had a lower SUVmax compared with patients with wild-type EGFR (63 vs. 40%) (Table I).
Association of SUV<sub>max</sub> and EGFR mutations
Using χ2 analysis, the EGFR mutation status was identified to be significantly associated with sex, smoking status, pathological type and the SUVmax of the primary tumor (Table I). The potential association between SUVmax and EGFR mutation was investigated, and it was identified that the SUVmax was significantly lower in patients with EGFR mutations (mean, 6.52±0.38) compared with that in patients with wild-type EGFR (mean, 9.37±0.31; P<0.001) (Fig. 2C). ROC curve analysis revealed an SUVmax cut-off point of 7.8 (Table I), with an AUC of 0.75 (95% confidence interval, 0.68-0.83; Fig. 2D). Using χ2 analysis, EGFR mutation status was identified to be significantly associated with sex, smoking status, tumor histopathology and SUVmax of the primary tumor (Table I). Using multivariate analysis, smoking status (never-smoking), histopathology (adenocarcinoma) and SUVmax (≤9.91) were the statistically significant predictors of EGFR mutations (Table II).
Patients were divided into two groups according to this threshold and it was identified that EGFR mutations were more frequent in patients with a low SUVmax (≤9.92) compared with in patients with a high SUVmax (>9.92) (53 vs. 1.8%; P<0.001).
GLUT1 expression is downregulated in EGFR mutated cell lines
Since GLUT1 has been investigated as an important regulator of glucose transport, we hypothesized that the decreased SUVmax associated with EGFR mutations may be caused by downregulated GLUT1 expression. RT-qPCR revealed that GLUT1 mRNA was decreased in the PC-9 and NCI-H1975 cell lines compared with in the A549 cell line (0.82±0.07 and 0.72±0.04 vs. 0.98±0.04; P<0.05; Fig. 3A), indicating that decreased GLUT1 may be involved in the downregulated FDG uptake in patients with an EGFR mutated status.
Decreased ROS activity is detected in the PC-9 cell lines
Previous studies have identified that intracellular ROS serve important functions in regulating GLUT1 expression. To determine whether the different GLUT1 expression levels in A549, PC-9 and NCI-H1975 cells are influenced by ROS levels, the intracellular ROS level was determined in A549, PC-9 and NCI-H1975 cells. As presented in Fig. 3B, a marked decrease in the intracellular concentration of ROS was identified in PC-9 and NCI-H1975 cells, which confirmed our hypothesis.
NOX4 mRNA and protein levels are decreased in EGFR-mutated cell lines
NOX4 is a gene that maps to the 11q14.3 region and its sequence has been strictly conserved throughout evolution. The NOX4 gene consists of 29 exons, and the NOX4 protein consists of 578 amino acids. This gene also encodes a member of the NOX4 family of enzymes that functions as the catalytic subunit of the NADPH oxidase complex (Fig. 3C and D). Previous studies have also identified that NOX4 serves crucial functions in ROS production (18,22). To investigate whether the altered ROS activity was influenced by the NOX4 molecule, mRNA and protein expression levels of NOX4 were determined in the A549, PC-9 and NCI-H1975 cell lines. The NOX4 mRNA was decreased by 20% in PC-9 (P<0.01) and by 14% in NCI-H1975 (P<0.05) cells, respectively (Fig. 3E), whereas the protein expression decreased by 13 and 16% in PC-9 and NCI-H1975 cells, respectively (both P<0.05), compared with the A549 cell line (Fig. 3F).
Discussion
In the present study, the association between SUVmax and EGFR mutation status was investigated in patients with NSCLC. The results revealed that patients who harbored an EGFR mutation exhibited decreased SUVmax values, and further studies revealed that the EGFR mutation alters the SUVmax partially via the NOX4/ROS/GLUT1 axis.
The aims of the present study were as follows: i) To determine whether tumor metabolism can add significant value for predicting EGFR gene mutation; and ii) to investigate the molecular mechanisms by which lung lesions alter the metabolic pathway.
The selection of a suitable therapeutic strategy for a patient suffering from lung cancer is based on the gene status, particularly EGFR. In 2009, Lara-Guerra et al (23) carried out a Phase II study that included 31 patients clinically diagnosed as stage I NSCLC, who received pre-operative gefitinib. The results indicated that tumor shrinkage was frequently seen in women who had never smoked, and the EGFR mutation was the strongest predictor of response. Apart from stage I patients, gefitinib is still useful in patients with stage III/IV NSCLC (3 achieved complete response, 13 exhibited partial response, 3 had stable disease and 2 were discontinued for side effects among the total 21 patients) (24). All these studies indicate the urgent requirement to validate the gene mutation, and a less invasive test method is desirable. Although individual gene detection has been recommended for advanced NSCLC, certain problems (including tumor inaccessibility, insufficient sample tissue for detection and unwillingness to perform invasive detection) have hindered this potential benefit for patients with advanced NSCLC (25). Consequently, a non-invasive strategy for predicting EGFR gene mutation status is advantageous, and the SUVmax, which represents the most active metabolic location within the lesion, has been used as the most convenient metabolic parameter in malignant diseases including lung cancer. However, the association between EGFR mutation and SUVmax differs markedly among studies, and the data from previous association studies are summarized in Table III. These differences are observed because, first, the SUVmax, a semi-quantitative index, varies with different PET scanners, fasting durations, plasma levels and region of interest parameters, and, secondly, different studies enrolled various sample sizes and disparate pathology types, which may also contribute to variation. A systematic meta-analysis should be performed to evaluate these results (26). The results of the present study indicated that never-smoking, female and lower SUVmax were the most significant predictive factors for the presence of the EGFR mutation, in accordance with previous studies (26,27). Using a patient’s clinicopathological and imaging data, which represents the non-invasive examination, to diagnose EGFR mutation status and other mutations is of marked importance. On the basis of the results of the present study, with an SUVmax cut-off value of 9.92, the sensitivity and specificity for our prediction model were 98.15 and 53.85%, respectively.
On the basis of the result that decreased FDG uptake was identified in patients harboring an EGFR mutation (9.37±0.31 vs. 6.52±0.38, wild-type vs. mutation), the ROC curve was first analyzed, and it was identified that the SUVmax cut-off point was 9.92 and the AUC was 0.75 (95% confidence interval, 0.68-0.83). Next, we hypothesized that GLUT1, which serves important functions in transporting glucose and is expressed during all stages of embryonic development (35), may function as a key molecule in regulating FDG uptake. Western blotting revealed that GLUT1 decreased markedly in PC-9 and NCI-H1975 cells compared with in A549 cells, indicating that EGFR mutation status may regulate FDG uptake by altering GLUT1 expression. Previous studies have identified that GLUT1 expression may be regulated by ROS in disparate pathways. Under normal conditions, ROS can be produced as a product of normal mitochondrial energy metabolism, and slightly increased ROS functions as a molecular signal to activate various signaling pathways including glucose uptake. However, a persistently high ROS level may reverse the traditional signaling pathway (36). In the present study, the ROS level was determined using DHE and it was revealed that decreased ROS activity was detected in PC-9 and NCI-H1975 cell lines, which is consistent with previous studies. Fiorentini et al (37) identified that decreasing ROS activity by adding the antioxidant EUK-134 downregulated total GLUT1 expression, partially indicating a positive correlation between ROS activity and GLUT1 expression. The dysregulation of the redox balance in cancer cells exerts crucial functions in tumor development and the response to anticancer therapies (38). Kawano et al (39) transfected 293T cells with a vector expressing an Ex19del mutant of human EGFR and identified a marked increase in the intracellular concentration of ROS, indicating a potential association between EGFR mutation and ROS activity. In the present study, ROS activity was also determined, and it was identified that PC-9 cells and NCI-H1975 cells expressed lower ROS levels.
Previous studies have also identified that NOX4 serves crucial functions in ROS production (18,22). NOX4 is a gene that maps to the 11q14.3 region, and its sequence has been strictly conserved throughout evolution. The NOX4 gene consists of 29 exons, and the NOX4 protein consists of 578 amino acids. This gene encodes a member of the NOX family of enzymes that functions as the catalytic subunit of the NADPH oxidase complex. The encoded protein is localized to non-phagocytic cells where it acts as an oxygen sensor and catalyzes the reduction of molecular oxygen to various ROS. The ROS generated by this protein have been implicated in numerous biological functions including signal transduction, cell differentiation and tumor cell growth (40,41). Furthermore, Prata et al (16) identified that NOX4-derived ROS could maintain a high glucose uptake rate by upregulating GLUT1 in a leukemic cell line (16). In the present study, it was identified that NOX4 mRNA and protein levels were significantly decreased in PC-9 and NCI-H1975 cells, compared with in A549 cells, suggesting an underlying molecular mechanism by which ROS activity is decreased. Previous studies have identified that increased ROS activates hypoxia-inducible factor α (HIF-α) and bind to HIF-α-response elements in the promoter regions of target genes (including GLUT1), thereby increasing GLUT1 mRNA and protein levels (42,43). Conversely, in patients with NSCLC harboring an EGFR mutation, inhibited ROS activity may be responsible for the downregulated GLUT1 protein level (Fig. 4). Indeed, it has been identified previously that NOX4 is essential for EGFR TKI activity. Orcutt et al (44) revealed that the cytotoxicity of erlotinib, an EGFR TKI, was mediated by induction of oxidative stress by inducing the expression of NOX4 in human head and neck cancer. Sobhakumari et al (45) also revealed that erlotinib increased NOX4 mRNA and protein expression by increasing its promoter activity and mRNA stability in FaDu cells, which potentially implied that the primary NOX4 expression is not enough for erlotinib function and the relatively decreased NOX4 expression possibly be a trigger which activates the erlotinib activity.
The limitations of the present study should be clarified. First, the study was designed retrospectively, with a relatively small size (previous studies have ranged in size between 34 and 734 patients). With the accumulation of these small-sample studies, a relatively objective and correct conclusion or opinion may be drawn, for example, by meta-analysis. In addition, a particular geographical issue should be considered. The patient cohort in the present study was primarily from Yunnan Province, an undeveloped, secluded and mountainous province of southwestern China, therefore a number of individuals in this region are unable to afford the relatively expensive cost of PET-CT and gene mutation detection, directly leading to the small sample size. Secondly, a bias could have existed in the process of the patient selection process since the majority of the patients resided in Yunnan Province that is known for high lung cancer rates (46-48). Thirdly, differences in metabolic parameters among different EGFR mutations, and between EGFR mutation and other important mutations (e.g. KRAS) were not discussed, which we intend to address in future studies. In the present study, although direct evidence remains limited, patients with EGFR mutation exhibited obviously decreased SUVmax compared with those with no EGFR mutation. χ2 analysis revealed that SUVmax is one predictor of EGFR mutation status and univariate analysis indicated that SUVmax was the only predictor of EGFR mutation. In the future, with the requisite equipment, FDG uptake among different lung cancer cells with various EGFR mutation status may be detected, which will provide direct evidence. The sample size will be increased and follow-up of patients assessed in the present study will be continued, and it is intended to publish survival results in the future. Finally, it should also be recog-nized that tissue testing is the gold standard for judging EGFR mutation status.
In conclusion, the results of the present study from clinical samples and cell lines indicate that the FDG uptake was decreased in patients with NSCLC with EGFR mutation. In addition, with a cut-off value of 9.92, the SUVmax is useful in predicting EGFR mutation, indicating that PET-CT may be a useful non-invasive instrument for predicting EGFR mutation in patients with NSCLC, thereby optimizing the clinical treatment strategy. In addition, further experiments at the cell and molecular levels validated that the NOX4/ROS/GLUT1 axis is responsible for decreased FDG uptake in patients with NSCLC with EGFR mutation, which may reveal potential treatment targets.