Abstract
Pharmacologic treatment of acromegaly is currently based upon assay-error strategy, the first-generation somatostatin receptor ligands (SRL) being the first-line treatment. However, about 50% of patients do not respond adequately to SRL. Our objective was to evaluate the potential usefulness of different molecular markers as predictors of response to SRL. We used somatotropinoma tissue obtained after surgery from a national cohort of 100 acromegalic patients. Seventy-one patients were treated with SRL during at least 6 months under maximal therapeutic doses according to IGF1 values. We analyzed the expression of SSTR2, SSTR5, AIP, CDH1 (E-cadherin), MKI67 (Ki-67), KLK10, DRD2, ARRB1, GHRL, In1-Ghrelin, PLAGL1 and PEBP1 (RKIP) by RT-qPCR and mutations in GNAS gene by Sanger sequencing. The response to SRL was categorized as complete response (CR), partial (PR) or non-response (NR) if IGF1 was normal, between >2<3 SDS or >3 SDS IGF1 at 6 months of follow-up, respectively. From the 71 patients treated, there were 27 CR (38%), 18 PR (25%) and 26 NR (37%). SSTR2, Ki-67 and E-cadherin were associated with SRL response (P < 0.03, P < 0.01 and P < 0.003, respectively). E-cadherin was the best discriminator for response prediction (AUC = 0.74, P < 0.02, PPV of 83.7%, NPV of 72.6%), which was validated at protein level. SSTR5 expression was higher in patients pre-treated with SRL before surgery. We conclude that somatotropinomas showed heterogeneity in the expression of genes associated with SRL response. E-cadherin was the best molecular predictor of response to SRL. Thus, the inclusion of E-cadherin in subsequent treatment-decision after surgical failure may be useful in acromegaly.
Introduction
Acromegaly is a rare disease, usually diagnosed late in the disease evolution, being caused in most of the cases by a pituitary adenoma (Melmed 2006). Current therapeutic modalities include surgery, pharmacologic compounds and radiotherapy when the previous two options fail (Katznelson et al. 2014, Melmed et al. 2018). The disease is severe enough to try to diagnose it as early as possible as well as to cure or control hormonal hypersecretion quickly after diagnosis. Unfortunately, most of the patients do not reach both diagnosis and disease control early (Reid et al. 2010, Rochette et al. 2016); moreover, somatotropinomas are heterogeneous tumors that may show diverse responses to the different pharmacologic options currently available. The first-generation somatostatin receptor ligands (SRL), octreotide and lanreotide, are the compounds recommended as first-line option when pharmacologic treatment is considered. When this treatment fails, which occurs in about 50% of cases (varying depending on the series from 30 to 75% (Freda 2002, Gadelha et al. 2017)), due mostly, but not exclusively, to the different expression levels of somatostatin receptors (SSTR) in the somatotropinoma surface (Wildemberg et al. 2013, Venegas-Moreno et al. 2018), second generation SRL or the GH receptor antagonist pegvisomant are used for controlling hormonal hypersecretion (Melmed et al. 2018). Current guidelines do not indicate which pharmacologic option best fits for an individual case, as treatment-decision process is based upon the trial-error approach (Chanson & Salenave 2008, Gadelha 2015). As a consequence, attainment of disease control may last considerably for a given patient after diagnosis. Additionally, response to first-generation SRL may be positive but incomplete, requiring not just the substitution by another compound but sometimes the combination with another drug, such as pegvisomant or dopamine agonists.
In the last few years, different studies (Gadelha et al. 2017, Gatto et al. 2017) have shown that the therapeutic response to first-generation SRL may be explained by the expression of the different SSTRs as well as by the downstream signaling in the somatotropinoma cells. Accordingly, the expression of SSTR2 and SSTR5 (Casar-Borota et al. 2013, Gatto et al. 2013, Wildemberg et al. 2013, Venegas-Moreno et al. 2018), the SSTR2/SSTR5 ratio (Taboada et al. 2007), the expression of the long and short DRD2 isoforms (Venegas-Moreno et al. 2018), aryl hydrocarbon receptor interacting protein (AIP) (Chahal et al. 2012), E-cadherin (CDH1) (Fougner et al. 2010, Kiseljak-Vassiliades et al. 2015a), Ki-67 (MKI67) (Kasuki et al. 2013), arrestin beta-1 (ARRB1) (Gatto et al. 2016, Coelho et al. 2018), ghrelin (GHRL), intron 1 ghrelin (In1-ghrelin) (Ibáñez-Costa et al. 2015), ZAC1 (PLAGL1) (Chahal et al. 2012), Raf kinase inhibitory protein (PEBP1 or RKIP) (Fougner et al. 2008), kallikrein 10(KLK10) (Rotondo et al. 2015) and mutations in guanine nucleotide binding protein (G Protein), alpha stimulating activity polypeptide 1 (GNAS or GSP) mutations (Efstathiadou et al. 2015) have been involved in SRL response. In most of the studies, each of these molecules has been evaluated individually and their ability to predict therapeutic response has been variable. We aimed to evaluate all these potential markers to SRL response in a large series of acromegaly patients in which these compounds were used in order to identify those markers having a high predictive capacity.
Patients and methods
Patients
A transnational cohort consisting of 100 acromegaly patients from 26 tertiary centers from all over Spain who had undergone pituitary surgery and had tissue availability (RNA later preserved tumor sample) were included in the study. In those patients in which more than one surgery was performed, only one sample tumor per patient was analyzed. The description of the phenotypic characteristics of the cohort is presented in Table 1. The heterogeneous cohort of patients tries to reflect the daily practice of acromegaly management. Basal molecular and clinical information for correlation analyses were available in the whole cohort. Of these 100 patients, 67 had received SRL treatment (octreotide or lanreotide) before surgery and 33 had not received treatment before surgery. Follow-up data after surgery regarding SRL response were available in 71 patients, including 51 out of 67 cases (51% females, mean age 45.3 ± 13 years) who had received SRL treatment before surgery and 20 out of 33 patients who had not (51% females, mean age 44.6 ± 13 years). All patients in which clinical information was available at follow-up and were treated after surgery for at least 6 months under maximal effective therapeutic (octreotide or lanreotide) doses according to IGF1 values were included in the analysis; this was possible in 71 cases. In the 29 remaining patients, 22 were cured after surgery and 7 were lost to follow-up. The IGF1 levels reported, evaluating IGF1 reduction and categorization of response to SRL, correspond only to postsurgical follow-up. Patients were categorized according to therapeutic response to SRL as complete response (CR), partial (PR) or non-responders (NR) if IGF1 was normal, between >2<3 SDS or >3 SDS IGF1 at 6 months of follow-up, respectively.
General and clinical characteristics of the patients and the tumors included in the study.
| Patients characteristics | |
|---|---|
| Cohort (n) | 100 |
| Male/female | 44/56 |
| Age | 45.5 ± 13.28 |
| Medical treatment | |
| DA treated | 12 |
| SRL presurgery | 67 |
| SRL response | |
| Non-responders | 26 |
| Partial responders | 18 |
| Complete responders | 27 |
| Comorbidities (%) | |
| Diabetes | 27 |
| HBP | 29 |
| Dyslipidemia | 27 |
| Cancer | 6 |
| Cerebrovascular accident | 3 |
| Cardiovascular incident | 13 |
| Visual alterations | 18 |
| Tumor characteristics (%) | |
| Macroadenoma | 79 |
| Extrasellar growth | 77 |
| Sinus invasion | 61 |
Follow up data after surgery regarding SRL response with a minimum follow-up of 6 months were available in 71 patients, 51 who had received SRL treatment before surgery and 20 patients who had not.
Accepting an alpha risk of 0.05 and a beta risk of 0.2 in a two-sided test, our cohort of subjects is able to recognize as statistically significant a minimum difference of 0.38 units between any pair of groups assuming that three groups exist. The common deviation is assumed to be 0.4.
The study was conducted in accordance with the ethical principles of the Declaration of Helsinki and implemented and reported in accordance with the International Conference on Harmonised Tripartite Guideline for Good Clinical Practice. The study was approved by the Germans Trias i Pujol Hospital Ethical Committee for Clinical Research (Ref.: EO-11-080, http://www.ceicgermanstrias.cat/index.html). The protocol and informed consent forms were approved by the institutional review board of the participating centers, independent ethics committee and/or research ethics board of each study site. All patients provided written informed consent to participate in the trial.
RNA isolation and reverse transcription
Total RNA was isolated from pituitary adenomas using AllPrep DNA/RNA/miRNA Universal Kit (Qiagen). The quantity and purity of extracted RNA was quantified by measuring optical density at 260 and 280 nm using NanoDrop™ 1000 Spectrophotometer (RRID:SCR_016517, Thermo Fisher Scientific).
Integrity of the RNA was checked by agarose gel electrophoresis. To remove contaminating genomic DNA, samples were treated with RNAse-free DNAse twice, during the RNA extraction procedure following the manufacturer’s protocol and before the retrotranscription using ezDNase Enzyme (Invitrogen).
Five hundred nanograms of total RNA were reverse transcribed using SuperScript IV reverse transcriptase (Invitrogen) and random hexamers in a final volume of 20 µL according to the manufacturer’s protocol.
Quantitative polymerase chain reaction
Gene expression was quantified using Taqman assays (Applied Biosystems). The genes analyzed were somastostatin receptor 2 (SSTR2, Hs00990356_m1), somastostatin receptor 5 (SSTR5, Hs00990408_s1), short dopamine receptor D2 (DRD2) isoform (Hs01014210_m1), long DRD2 isoform (Hs01024460_m1), arrestin beta 1 (ARRB1, Hs00930516_m1), pleiomorphic adenoma gene-like 1 (PLAGL1, Hs00414677_m1), Raf kinase inhibitory protein (RKIP/PEBP1, Hs01110783_g1), E-cadherin (CDH1, Hs01023894_m1), Ki-67 (MKI67, Hs01032443_m1), ghrelin and obestatin prepropeptide (GHRL, Hs01074053_m1), aryl hydrocarbon receptor interacting protein (AIP, Hs00610222_m1) and a custom assay was ordered for intron 1 ghrelin In1-GHRL (AJ89KWC). We tested six reference genes to normalize gene expression: hypoxanthine phosphoribosyltransferase 1 (HPRT1, Hs99999909_m1), proteasome 26S subunit ATPase 4 (PSMC4, Hs00197826_m1), glucuronidase beta (GUSB, Hs00939627_m1), TATA-box binding protein (TBP, Hs00427621_m1), mitochondrial ribosomal protein L19 (MRPL19, Hs01040217_m1) and phosphoglycerate kinase 1 (PGK1, Hs00943178_g1) and selected the last three reference genes based on their stability in our samples according to Chainy software (available on: http://maplab.imppc.org/chainy/) (Mallona et al. 2017).
Quantitative polymerase chain reactions (qPCR) were carried out in a 7900HT Fast Real-Time PCR System (Applied Biosystems). We used TaqMan Gene Expression Master Mix (Applied Biosystems), and the amplification reactions were performed in triplicate for each sample in a final volume of 10 µL in 384-well plates. To minimize the inter-assay variation, all the genes, including the reference genes, for each sample were analyzed in the same plate. The relative quantification to reference genes was calculated according to geNorm (RRID:SCR_006763, https://genorm.cmgg.be/) algorithms (Vandesompele et al. 2002).
GNAS sequencing
The mutations in guanine nucleotide binding protein (G protein), alpha stimulating activity polypeptide 1 (GNAS, also known as GSP oncogene) were screened by Sanger sequencing (GATC Biotech, Cologne, Germany). Samples were analyzed for mutations at codons 201 and 227 in exons 8 and 9, respectively, using cDNA and the primers 5′-CAAGCAGGCTGACTATGTGCCGA-3′ (forward) and 5′-CCACCACGAAGATGATGGCAGTC-3′ (reverse).
Biochemical and hormonal assays
After an overnight fast, blood samples were collected from patients at baseline and at different follow-up times. Serum IGF1 was measured by two different methods and normalized for comparisons by expressing SDS values. Method 1, a two-site IRMA (Immunotech IGF1 kit; Immunotech-Beckman, Marseille, France). Expected values depending on age were: 20–30 years, 220–550 ng/mL; 30–40 years, 140–380 ng/mL; 40–50 years, 54–330 ng/mL; and 50–60 years, 94–285 ng/mL. Intra-assay CV was less than 6.3%; inter-assay CV, 6.8%; and sensitivity, 30 ng/mL. Method 2 was a non-extraction IRMA (Diagnostic Systems Laboratories, Webster, TX, USA). The theoretical sensitivity, or minimum detection limit, calculated by interpolation of the mean plus two s.d. values of 20 replicates of the 0 ng/mL IGF1 standard was 2 ng/mL. The inter-assay CV was 7.4 and 4.2, respectively, for the concentrations 32.5 and 383.8 ng/ml. The inter-assay CV was 7 and 3.9, respectively, for the mean concentration values 34.03 and 373.86 ng/mL.
To better categorize the disease status and the response to SRL and to avoid the variability over time of IGF1 measurement, results are expressed as SDS according to sex and age and percentage of decrease over basal value. Therefore, IGF1 greater than 3 SDS was considered not responsive to SRL treatment, between 2 and 3 SDS was considered a partial response to SRL and less than 2 SDS was considered a complete response to SRL treatment (Puig-Domingo et al. 2010).
Immunohistochemistry
Those markers that performed better while analysed by gene expression were subsequently evaluated at protein level by immunohistochemistry. Thus, forty-six somatotropinoma tissue samples were available for immunostaining of E-cadherin, SSTR2a, Ki-67 and cytokeratin CAM 5.2 as well, as CAM 5.2 has previously demonstrated to identify accurately densely granulated and sparsely granulated somatotropinomas with good identification power of responsiveness and non-responsiveness to SRL, respectively (Al-Brahim & Asa 2006, Bakhtiar et al. 2010). Formalin-fixed paraffin-embedded tumour samples were cut into sequential 4-µm-thick sections and stained using a fully automated Ventana BenchMark ULTRA stainer (Ventana, Tucson, AZ, USA) according to the manufacturer’s instructions. Binding of peroxidase-coupled antibodies was detected using diaminobenzidine as a substrate, and the sections were counterstained with hematoxylin.
The mouse monoclonal anti-cytokeratin antibody and the mouse monoclonal anti-E-cadherin antibody (Ventana) were purchased as prediluted antibodies, with a concentration of 11 µg/dL and 0.314 µg/dL, respectively. The rabbit monoclonal anti-SSTR2a antibody (clone UMB-1, Abcam) was used at a dilution of 1:100. To analyze Ki-67, we used the rabbit monoclonal anti–Ki67 antibody 30-9 (ready-to-use formulation; Ventana). Normal appendix tissue served as the positive control for CAM 5.2 staining and mammary invasive ductal carcinoma for E-cadherin staining.
Immunostaining for E-cadherin was scored in three intensities (0: negative, 1+: weak positivity, 2+: strong positivity) and for each intensity, the percentage of cells was determined. For the classification of the intensities, we considered 0 (negative) when there were no positivity; 1+ (weak positivity) when the adenoma cells seemed negative at low magnification (×40) but were truly positive at high magnification (×200) and we considered 2+ when the adenoma cells were clearly positive at low magnification (×40). We calculated an IHC score multiplying the percentage of cells of each intensity by the score intensity (0–200). Loss of E-cadherin was considered for IHC scores equal to 0. Partial loss of E-cadherin was considered for IHC scores below 100.
Immunostaining for SSTR2 was scored using an H-score as performed in Franck et al. (2017). First, membrane and cytoplasmic staining intensity (0: no staining, 1+: weak positivity, 2+: moderate positivity and 3+: strong positivity) was determined for each field and then the percentage of cells at each staining intensity level was calculated. An H-score was assigned using the following formula: (1 × (% cells 1+) + 2 × (% cells 2+) + 3 × (% cells 3+)).
Ki-67 score was expressed as the percentage of the number of immunostained nuclei among the total number of nuclei of tumor cells regardless of the immunostaining intensity. The counting was performed in three randomly selected fields of the adenoma tissue section at ×400 magnification.
For the CAM 5.2 staining, the adenomas were classified in two groups: dot-type (when the pattern was exclusively dot-type which identifies accurately sparsely granulated somatotropinomas) and not-only-dot-type (when there were other patterns in addition or not to the dot-type pattern which identifies accurately densely granulated somatotropinomas).
Statistical analysis
Descriptive results were expressed as mean ± s.d. or median and 25th to 75th percentiles, as appropriate. Unsupervised hierarchical clustering was used to investigate the potential identification of patient’s response subgroups, based on their molecular expression profile. Spearman or Pearson bivariate correlations were performed for all quantitative variables and differences between groups were compared using comparison of mean tests (Student’s t-test, Wilcoxon signed-rank test and Kruskal–Wallis and ANOVA as appropriate). A multinomial logistic regression model was used to determine the differences in each normalized gene expression between complete response and resistant patients. The model was adjusted by age, gender and SRL pre-surgical treatment. Receiver operating characteristic (ROC) curve analyses were performed to assess the classification power of each logistic regression model. Samples from all groups within an experiment were processed at the same time. The P values were two-sided, and statistical significance was considered when P < 0.05. All statistical analyses were performed using STATA (StataCorp LLC, College Station, Texas, USA, RRID:SCR_012763) and R version 3.3.2 (R Project for Statistical Computing, RRID:SCR_001905). Unsupervised hierarchical clustering was performed using the R package pheatmap (Pretty Heatmaps, https://CRAN.R-project.org/package=pheatmap). The graphical representation was done using package ggplot 2 (RRID:SCR_014601, Whickham https://CRAN.R-project.org/package=ggplot2) and the P values were added using ggpubr package (‘ggplot2’ Based Publication Ready Plots, https://CRAN.R-project.org/package=ggpubr). Finally, the ROC curves were plotted using pROC package (Display and Analyze ROC Curves, https://CRAN.R-project.org/package=pROC).
Results
Clinical variables according to biomarkers expression
In the whole cohort (n = 100), we analyzed the expression of 12 genes previously reported to be involved in SRL response, including SSTR2, SSTR5, AIP, E-cadherin, Ki-67, KLK10, DRD2, ARRB1, GHRL, In1-Ghrelin, PLAGL1 and RKIP. Tumor size was related to SSTR2 (Pearson’s r = 0.25, P = 0.01) and showed a negative association with DRD2 (short DR2D isoform Pearson’s r = −0.29, P < 0.01, and long DRD2 isoform Pearson’s r = −0.37, P < 0.001) and E-cadherin (Pearson’s r = −0.28, P < 0.01). Extrasellar extension was also related to long DRD2 isoform (P = 0.01) and Ki-67 (P = 0.04). Moreover, visual alteration was negatively related to DRD2 (P = 0.01 for both isoforms) and E-cadherin (P = 0.02) (Fig. 1). We also found a negative correlation between IGF1 levels at diagnosis and expression of ARRB1 (Pearson’s r = −0.31, P = 0.002), KLK10 (Pearson’s r = −0.23, P = 0.02) and E-cadherin (Pearson’s r = −0.29, P = 0.003). Furthermore, we analyzed the correlation of the expression of each marker with IGF1 index at diagnosis and IGF1 % decrease after SRL treatment, and E-cadherin was the only marker that showed significant correlations with the three IGF1-related measurements (Supplementary Table 1, see section on supplementary materials given at the end of this article), while Ki-67 has the strongest correlation with IGF1 % decrease (Pearson’s r = −0.357, P = 0.002).
Boxplot showing gene expression according to tumor characteristics. Relative expression in tumors smaller and larger than 2 cm (an arbitrary threshold that separates our cohort in two equivalents subsets) (A, B and C), in tumors causing visual alterations before the surgery (D, E and F) and in tumors with or without extrasellar extension (G and H). A full color version of this figure is available at https://doi.org/10.1530/ERC-18-0565.
Citation: Endocrine-Related Cancer 27, 6; 10.1530/ERC-18-0565
Boxplot showing gene expression according to tumor characteristics. Relative expression in tumors smaller and larger than 2 cm (an arbitrary threshold that separates our cohort in two equivalents subsets) (A, B and C), in tumors causing visual alterations before the surgery (D, E and F) and in tumors with or without extrasellar extension (G and H). A full color version of this figure is available at https://doi.org/10.1530/ERC-18-0565.
Citation: Endocrine-Related Cancer 27, 6; 10.1530/ERC-18-0565
Boxplot showing gene expression according to tumor characteristics. Relative expression in tumors smaller and larger than 2 cm (an arbitrary threshold that separates our cohort in two equivalents subsets) (A, B and C), in tumors causing visual alterations before the surgery (D, E and F) and in tumors with or without extrasellar extension (G and H). A full color version of this figure is available at https://doi.org/10.1530/ERC-18-0565.
Citation: Endocrine-Related Cancer 27, 6; 10.1530/ERC-18-0565
According to SRL biochemical categorized response analyzed in 71 patients, 27 patients (38%) were CR, 18 (25%) PR and 26 (37%) were considered NR. In 20 of these 71 cases, treatment with SRL was only given after surgical procedure, while the rest received SRL therapy before and after surgery. When an unsupervised hierarchical clustering analysis of the expression of the studied genes was performed in all 71 cases, we found that clustering was not related to or influenced by either the overall SRL response or the SRL treatment given before or after surgery (Fig. 2). This indicates that, as a group, acromegaly patients treated with SRL do not present a specific pattern of expression in relation to a given response to SRL and, thus, confirming the heterogeneous nature of somatotropinomas.
Dendrogram and unsupervised hierarchical clustering heat map of the expression of analyzed SRL response biomarkers using Ward’s minimum variance method and Minkowski distance. For every patient, GNAS mutation, SRL treatment before surgery and SRL response category are shown if available (n = 71).
Citation: Endocrine-Related Cancer 27, 6; 10.1530/ERC-18-0565
Dendrogram and unsupervised hierarchical clustering heat map of the expression of analyzed SRL response biomarkers using Ward’s minimum variance method and Minkowski distance. For every patient, GNAS mutation, SRL treatment before surgery and SRL response category are shown if available (n = 71).
Citation: Endocrine-Related Cancer 27, 6; 10.1530/ERC-18-0565
Dendrogram and unsupervised hierarchical clustering heat map of the expression of analyzed SRL response biomarkers using Ward’s minimum variance method and Minkowski distance. For every patient, GNAS mutation, SRL treatment before surgery and SRL response category are shown if available (n = 71).
Citation: Endocrine-Related Cancer 27, 6; 10.1530/ERC-18-0565
GNAS mutation analysis
GNAS mutations were studied in a subset of 50 patients and we found mutations in 33%, c.601C>T being the most frequent (Fig. 3). SRL response was not significantly different in those patients presenting GNAS mutations; mutated cases were found in 29% of the CR group, 38% of PR and 36% of NR. Neither did we find that clinical variables were related to mutational status regarding comorbidities, tumor size and age among the patients in which the analysis was performed, nor any association with the expression of the different analyzed markers with GNAS mutations (Fig. 2).
Results from mutational analysis of GNAS gene (n = 50). (A) Percentage of the different mutations found in our cohort. (B) Proportion of patients carrying GNAS mutations grouped according to therapeutic response to SRL in complete responders (CR), partial responders (PR) and non-responders (NR). A full color version of this figure is available at https://doi.org/10.1530/ERC-18-0565.
Citation: Endocrine-Related Cancer 27, 6; 10.1530/ERC-18-0565
Results from mutational analysis of GNAS gene (n = 50). (A) Percentage of the different mutations found in our cohort. (B) Proportion of patients carrying GNAS mutations grouped according to therapeutic response to SRL in complete responders (CR), partial responders (PR) and non-responders (NR). A full color version of this figure is available at https://doi.org/10.1530/ERC-18-0565.
Citation: Endocrine-Related Cancer 27, 6; 10.1530/ERC-18-0565
Results from mutational analysis of GNAS gene (n = 50). (A) Percentage of the different mutations found in our cohort. (B) Proportion of patients carrying GNAS mutations grouped according to therapeutic response to SRL in complete responders (CR), partial responders (PR) and non-responders (NR). A full color version of this figure is available at https://doi.org/10.1530/ERC-18-0565.
Citation: Endocrine-Related Cancer 27, 6; 10.1530/ERC-18-0565
Influence of SRL treatment given before or after surgery in the expression of molecular markers
Molecular markers expression was compared between patients who had received SRL treatment before surgery (n = 67) and those that had not received it (n = 33). We found that those in which pre-surgical treatment was performed showed higher expression levels of RKIP and SSTR5 (P = 0.006 and 0.017, respectively) than those not pre-treated (Fig. 4). Interestingly, the expression of the SSTR5 in the pre-treated patients was not different according to the SRL response (0.46 ± 0.61, 1.41 ± 2.39 and 0.51 ± 0.39, SSTR5 expression for CR, PR and NR, respectively, P = 0.087), suggesting that the mechanism regulating SSTR5 expression upon SRL treatment is different from that reducing GH secretion. By contrast, SSTR2 expression was not affected by pre-surgical treatment (P = 0.46) at mRNA level. We validated this result by SSTR2a immunohistochemistry (IHC) (P = 0.28).
Relative expression of SSTR2 (A), RKIP (B) and SSTR5 (C) in tumors receiving SRL or not receiving SRL before surgery (n = 100). A full color version of this figure is available at https://doi.org/10.1530/ERC-18-0565.
Citation: Endocrine-Related Cancer 27, 6; 10.1530/ERC-18-0565
Relative expression of SSTR2 (A), RKIP (B) and SSTR5 (C) in tumors receiving SRL or not receiving SRL before surgery (n = 100). A full color version of this figure is available at https://doi.org/10.1530/ERC-18-0565.
Citation: Endocrine-Related Cancer 27, 6; 10.1530/ERC-18-0565
Relative expression of SSTR2 (A), RKIP (B) and SSTR5 (C) in tumors receiving SRL or not receiving SRL before surgery (n = 100). A full color version of this figure is available at https://doi.org/10.1530/ERC-18-0565.
Citation: Endocrine-Related Cancer 27, 6; 10.1530/ERC-18-0565
Predictive response to SRL according to molecular markers expression
Neither SSTR5 nor SSTR5/SSTR2 ratio, ARRB1, PLAGL1, GHRL, In-1-Ghrelin and RKIP showed any statistically different expression among the three therapeutic response categories when the 71 cases were analyzed as a whole. AIP showed a trend toward significance when extreme phenotypes were compared (CR vs NR) with a P = 0.054 (Supplementary Table 1).
However, E-cadherin, SSTR2 and Ki-67 expression were associated with response to SRL (P = 0.006, P = 0.068 – near significance – and P = 0.03, respectively) (Fig. 5). Higher expression of E-cadherin and SSTR2 was observed in CR group when compared to NR (P < 0.003 and P < 0.03, respectively). The opposite pattern was observed for Ki-67, as NR showed higher levels (P < 0.001). Interestingly, E-cadherin and Ki-67 showed expression differences in a stepwise manner. E-cadherin was the marker that presented more differences between the three different categories of therapeutic response, showing a tendency between PR and NR (P < 0.1). E-cadherin presented 2.41-fold change between CR and NR and 1.52 when PR were compared to NR.
Relative expression of SSTR2 (A), Ki-67 (B) and E-cadherin (C) in complete responders (CR), partial responders (PR) and non-responders (NR) (n = 71). A full color version of this figure is available at https://doi.org/10.1530/ERC-18-0565.
Citation: Endocrine-Related Cancer 27, 6; 10.1530/ERC-18-0565
Relative expression of SSTR2 (A), Ki-67 (B) and E-cadherin (C) in complete responders (CR), partial responders (PR) and non-responders (NR) (n = 71). A full color version of this figure is available at https://doi.org/10.1530/ERC-18-0565.
Citation: Endocrine-Related Cancer 27, 6; 10.1530/ERC-18-0565
Relative expression of SSTR2 (A), Ki-67 (B) and E-cadherin (C) in complete responders (CR), partial responders (PR) and non-responders (NR) (n = 71). A full color version of this figure is available at https://doi.org/10.1530/ERC-18-0565.
Citation: Endocrine-Related Cancer 27, 6; 10.1530/ERC-18-0565
In addition, categorical analyses for each normalized gene expression in quintiles were performed to evaluate any nonlinearity in estimated effects. Interestingly, SSTR2 did not show any further risk increase over the second quintile. Similarly, E-cadherin expression levels did not increase the risk above the third quintile. This finding indicates the nonlinearity of gene expression for these two variables, suggesting that SRL response is related to a specific expression level conferring a permissive effect regarding therapeutic response closer to a categorical behavior of these biomarkers rather than a dose-response effect.
When multinomial logistic regression was constructed for extreme phenotypes (NR and CR), SSTR2 showed an AUC-ROC curve of 0.68, for a cut-off of 0.3, with a sensitivity of 61.5%, specificity of 69.2%, positive predictive value of 66.0% and negative predictive value of 62.6%; the OR for sensitivity toward response to SRL treatment was 3.729 (IC 97.5: 1.242–21.619; P = 0.06, non-significant). In contrast, ROC curve for E-cadherin showed an AUC of 0.74 and a sensitivity of 65.4%, specificity of 88%, positive predictive value of 83.7% and a negative predictive value of 72.6%. The effect sensitivity to SRL expressed as OR was 1.9319 (IC 97.25: 1.207–3.52; P < 0.02). When Ki-67 was analyzed by the multinomial logistic model, no significant results were obtained (P = 0.14). When the ROC curve was constructed combining both the expression of SSTR2 and E-cadherin together, no additional predictive power was obtained from the one observed for E-cadherin alone (P = 0.824) (Fig. 6).
ROC curves calculated with extreme phenotypes, complete responders (CR) (n = 27) and non-responders (NR) (n = 26) to SRL for E-cadherin (A) and SSTR2 (B). Comparison of the ROC curve obtained with E-cadherin expression alone or in combination with SSTR2 expression (C). A full color version of this figure is available at https://doi.org/10.1530/ERC-18-0565.
Citation: Endocrine-Related Cancer 27, 6; 10.1530/ERC-18-0565
ROC curves calculated with extreme phenotypes, complete responders (CR) (n = 27) and non-responders (NR) (n = 26) to SRL for E-cadherin (A) and SSTR2 (B). Comparison of the ROC curve obtained with E-cadherin expression alone or in combination with SSTR2 expression (C). A full color version of this figure is available at https://doi.org/10.1530/ERC-18-0565.
Citation: Endocrine-Related Cancer 27, 6; 10.1530/ERC-18-0565
ROC curves calculated with extreme phenotypes, complete responders (CR) (n = 27) and non-responders (NR) (n = 26) to SRL for E-cadherin (A) and SSTR2 (B). Comparison of the ROC curve obtained with E-cadherin expression alone or in combination with SSTR2 expression (C). A full color version of this figure is available at https://doi.org/10.1530/ERC-18-0565.
Citation: Endocrine-Related Cancer 27, 6; 10.1530/ERC-18-0565
In addition, gene expression correlations were explored to assess their possible relationships (Fig. 7). Interestingly, E-cadherin and SSTR2 had a moderate-strong positive correlation of r = 0.539 (P < 0.00001). Other correlations – either positive or negative – were also found between different biomarkers, indicating a multiplicity of functional relationship between them.
Spearman’s correlation matrix among the genes studied (n = 100). Genes are ordered according to hierarchical clustering using complete linkage method. Spearman’s correlation coefficients are shown in the matrix; the intensity of color reflects the correlation magnitude.
Citation: Endocrine-Related Cancer 27, 6; 10.1530/ERC-18-0565
Spearman’s correlation matrix among the genes studied (n = 100). Genes are ordered according to hierarchical clustering using complete linkage method. Spearman’s correlation coefficients are shown in the matrix; the intensity of color reflects the correlation magnitude.
Citation: Endocrine-Related Cancer 27, 6; 10.1530/ERC-18-0565
Spearman’s correlation matrix among the genes studied (n = 100). Genes are ordered according to hierarchical clustering using complete linkage method. Spearman’s correlation coefficients are shown in the matrix; the intensity of color reflects the correlation magnitude.
Citation: Endocrine-Related Cancer 27, 6; 10.1530/ERC-18-0565
Validation of E-cadherin expression by immunohistochemistry
We analyzed the protein amount of E-cadherin, SSTR2a and Ki-67 in 47 samples by IHC. E-cadherin H-score correlated with E-cadherin mRNA expression (Pearson’s r = 0.4, P < 0.003), and likewise E-cadherin H-score showed significant differences in SRL response stratification between CR and NR (P = 0.019) (Fig. 8A). Interestingly, E-cadherin loss by IHC defined as non-staining was found in PR and NR but not in CR (Fig. 8B). This behavior did not occur with partial loss which was found in both CR and NR. When multinomial logistic regression was constructed for extreme phenotypes (NR and CR), E-cadherin H-score showed an AUC-ROC curve of 0.79, for a cut-off of 30, with a sensitivity of 53.8%, specificity of 100%, positive predictive value of 100% and negative predictive value of 81.3% (Fig. 8C). These findings suggest that a completely negative IHC for E-cadherin may discard a complete biochemical control of IGF1 levels using only first-generation SRL.
Representative images of E-cadherin immunohistochemical scores in somatotropinomas (200×) (A). E-cadherin IHC score in complete responders (CR), partial responders (PR) and non-responders (NR) (n = 47) (B). E-cadherin IHC categorized in loss, partial loss and conserved in CR, PR and NR (C). ROC curve calculated with extreme phenotypes, complete responders (CR) (n = 13) and non-responders (NR) (n = 14), to SRL for E-cadherin IHC (D).
Citation: Endocrine-Related Cancer 27, 6; 10.1530/ERC-18-0565
Representative images of E-cadherin immunohistochemical scores in somatotropinomas (200×) (A). E-cadherin IHC score in complete responders (CR), partial responders (PR) and non-responders (NR) (n = 47) (B). E-cadherin IHC categorized in loss, partial loss and conserved in CR, PR and NR (C). ROC curve calculated with extreme phenotypes, complete responders (CR) (n = 13) and non-responders (NR) (n = 14), to SRL for E-cadherin IHC (D).
Citation: Endocrine-Related Cancer 27, 6; 10.1530/ERC-18-0565
Representative images of E-cadherin immunohistochemical scores in somatotropinomas (200×) (A). E-cadherin IHC score in complete responders (CR), partial responders (PR) and non-responders (NR) (n = 47) (B). E-cadherin IHC categorized in loss, partial loss and conserved in CR, PR and NR (C). ROC curve calculated with extreme phenotypes, complete responders (CR) (n = 13) and non-responders (NR) (n = 14), to SRL for E-cadherin IHC (D).
Citation: Endocrine-Related Cancer 27, 6; 10.1530/ERC-18-0565
SSTR2 H-score also showed a correlation with SSTR2 mRNA (Pearson’s r = 0.46, P < 0.01). However, the highest SSTR2 H-scores were found in the PR instead of the CR patients (Supplementary Fig. 1A). Furthermore, when multinomial logistic regression was constructed for CR vs NR and PR vs NR comparisons, SSTR2 showed an AUC-ROC curve of 0.62 (sensitivity of 50% and specificity of 77.8%) and an AUC-ROC curve of 0.70 (sensitivity of 60.2% and specificity of 76.2%), respectively, but neither of them were significant (P = 0.41 and P = 0.19, respectively).
Ki-67 IHC did not show any significant difference between the groups (Supplementary Fig. 1B). Moreover, the correlation between mRNA and protein was not significant (Pearson’s r = 0.21, P = 0.144). We think that the superior performance of the qPCR in comparison to IHC could be explained by the low levels of Ki-67 on these adenomas that make the levels difficult to quantify.
Influence of histological pattern on SRL response and E-cadherin expression
Finally, we analyzed the cytokeratin CAM 5.2 by IHC, as SRL response has been linked to histological subtypes (Kiseljak-Vassiliades et al. 2015b) and, particularly, CAM 5.2 immunostaining (Bakhtiar et al. 2010). Only 15 (32%) out of 47 samples presented a dot-type pattern. However, we observed that CR patients did not present dot-type tumors. Moreover, dot-type immunostaining for CAM 5.2 was negatively related to E-cadherin expression (Fig. 9). Altogether, these results suggest a link between the histological pattern, E-cadherin expression and SRL response in somatotropinomas.
Representative images of cytokeratin CAM 5.2 immunohistochemical patterns in somatotropinomas (200×) (A). Proportion of tumors with dot-type pattern and not-only-dot-type pattern according to therapeutic response to SRL (B). Relative expression of E-cadherin (C) and E-cadherin IHC score (D) in dot-type CAM 5.2 pattern and not-only-dot-type.
Citation: Endocrine-Related Cancer 27, 6; 10.1530/ERC-18-0565
Representative images of cytokeratin CAM 5.2 immunohistochemical patterns in somatotropinomas (200×) (A). Proportion of tumors with dot-type pattern and not-only-dot-type pattern according to therapeutic response to SRL (B). Relative expression of E-cadherin (C) and E-cadherin IHC score (D) in dot-type CAM 5.2 pattern and not-only-dot-type.
Citation: Endocrine-Related Cancer 27, 6; 10.1530/ERC-18-0565
Representative images of cytokeratin CAM 5.2 immunohistochemical patterns in somatotropinomas (200×) (A). Proportion of tumors with dot-type pattern and not-only-dot-type pattern according to therapeutic response to SRL (B). Relative expression of E-cadherin (C) and E-cadherin IHC score (D) in dot-type CAM 5.2 pattern and not-only-dot-type.
Citation: Endocrine-Related Cancer 27, 6; 10.1530/ERC-18-0565
Discussion
Acromegaly, a rare disease caused mostly by somatotropic tumors located in the pituitary gland, requires a prompt diagnosis in order to amend what often happens currently, where patients are only diagnosed after suspicion is aroused by the visibility of long-running phenotypic anatomical changes. If this was not already an important unresolved challenge, a second and equally necessary one is the need for an effective treatment for each patient to be delivered in a timely way. Different approaches have been taken to resolve this second issue, but they have not yet been introduced into clinical guidelines (Puig Domingo 2015, Kasuki et al. 2018). As somatotropinomas are heterogeneous tumors from a pathologic point of view, it is not unexpected that their response to first-line pharmacologic treatment, namely first-generation SRL, will not be homogeneous either. If certain biomarkers, such as T2 weighted signal from the MRI, seem to identify reasonably different response groups of patients (Puig-Domingo et al. 2010, Kiseljak-Vassiliades et al. 2015a), the information obtained from evaluation of somatotropinoma tissue samples after surgical treatment has raised high expectations, as nowadays the SRL signaling pathways are relatively well known. Therefore, different studies have been performed and published reporting that different receptors or molecules involved in SRL signaling may identify specific responses to these compounds (Gadelha et al. 2013). As far as we know, at least a dozen different biomarkers have been tested individually and have apparently shown interesting information potentially useful for developing a pharmacological treatment in those patients not cured by surgery. These molecules have never been considered altogether in the same patient and in a large series of patients in order to establish their true clinical usefulness for their possible incorporation into clinical guidelines as biomarkers of prediction of pharmacologic response.
In our study, we aimed to evaluate the mRNA expression of a combined panel composed of almost all SRL response biomarkers published in the last decade, for verification of previous results and definition of their predictive power, either individually or in specific combinations; those showing the best predictive performance were validated at protein level. We found that among all the biomarkers studied, E-cadherin, SSTR2 and Ki-67 showed potential usefulness for incorporation into clinical practice. E-cadherin expression was the best predictor between the SRL response categories. None of the other evaluated biomarkers showed statistical differences among the different response categories, although some of them showed a trend toward statistical significance, in particular AIP (P = 0.06). The SSTR2/SSTR5 ratio was not different among response categories and nor was PLAGL1 (ZAC1), a molecule which participates in the downstream pathway of SSTR2, in close relation to AIP (Chahal et al. 2012). Discrepancies with previous reports could be partially explained because of the methodology used and the population studied. We analyzed the gene expression of the panels of markers, while other studies measured the markers by IHC as in the case of AIP (Chahal et al. 2012) and KLK10 (Rotondo et al. 2015) or by Western blot as in the case of RKIP (Fougner et al. 2008). In the case of SSTR2, the concordance between RNA levels and IHC staining has been previously confirmed (Wildemberg et al. 2012), as we also found. However, in contrast to previous published results, we did not find a strong correlation between SSTR2 IHC and SRL response (Casar-Borota et al. 2013, Gatto et al. 2013, Wildemberg et al. 2013). This discrepancy could be explained by different reasons: (1) previous reports used a more homogeneous cohort (our patients were treated in different hospitals, some of them with octreotide while others with lanreotide); (2) response categories were different between studies and (3) scoring IHC is a subjective quantification of the protein that can vary between pathologists. The same reasons could also explain why other studies did not find a correlation between SSTR2 IHC and SRL response (Gonzalez et al. 2014). Regarding the discordance of SSTR2/SSTR5 ratio in our study with the work by Taboada (Taboada et al. 2007), it could be due to the fact that we used probe-base qPCR (Taqman technology) to measure gene expression, while the latter designed the primers and used intercalating dye-based qPCR which is less specific.
Of particular interest is the fact that the ROC curve analyses of E-cadherin and SSTR2 or their combination showed similar results, although E-cadherin presented better predictive power (either positive (84%) or negative (73%) for gene expression and even better for protein expression, positive predictive value of 100% and negative predictive value of 81.3%), and moreover, the combination of E-cadherin and SSTR2 was not superior than the one showed by E-cadherin alone. This indicates that if one single marker is to be chosen for incorporation into a decision-making therapeutic algorithm, E-cadherin might be the first one to be included to clinical guidelines.
Our study clearly exemplifies the biological heterogeneity of somatotropinomas, which by extension is also reflected in the response to SRL, the first-line pharmacological treatment of acromegaly recognized nowadays by clinical guidelines. Until now, all the studies investigating the biological behavior of somatotropinomas have built up a lot of information with high value for understanding somatotropinoma biology. Unfortunately, the expression of all the biomarkers identified so far is so wide and the variability among groups of responders and non-responders so high that it leads to an important degree of overlapping among SRL response categories, which does not allow the definition of specific cut-off values that could be currently applied to clinical practice. In this regard, E-cadherin expression is able to partially resist this overlapping effect among groups, although in some particular patients it may also fail because the overall predictive power – either positive or negative – is around 75% when gene expression is considered. The predictive value of E-cadherin levels was validated at protein level, which is of paramount importance from a clinical point of view, as IHC is easily implementable in the clinical routine.
Why is E-cadherin a better predictive biomarker than the rest when it would be more expected to find better results for the somatostatin receptor family? This issue requires further studies, although some remarkable information has already been generated by some studies, mostly coming from Bollerslev’s group. (Fougner et al. 2010, Lekva et al. 2012, 2013). E-cadherin is, among others, a biomarker of epithelial-mesenchymal transition (EMT), a biological process that seems also to be operative for somatotropinomas, at least in part, and may have implications for SRL response. As a matter of fact, the more advanced the EMT, the less responsive the tumor may be to SRL. This may explain in some way the biological heterogeneity shown in our cohort in which no specific expression pattern of the different markers evaluated present a strong concordance. The progressive loss of response to SRL seems to involve a concerted loss of E-cadherin and SSTR2 expression together with a gain in Ki-67, and thus the tumor losses its classic GH-secreting phenotype with a higher sparsely granulated pathologic pattern, according to cytokeratin CAM 5.2 staining (Fougner et al. 2012). Our results also validated that dot-type CAM 5.2 immunostaining correlates with poor response to SRL and E-cadherin loss in somatotropinomas (Bakhtiar et al. 2010).
Another interesting finding of our study is that SSTR5 expression was higher in those cases in which pre-surgical treatment with SRL was performed when compared to non-pre-treated patients. These patients were not different in terms of size of the tumor or other clinical variables, thus it is intriguing to understand this finding and it may be even postulated if SRL may have induced changes in the expression of SSTR5, a question that has previously been invocated for SSTR2 (Franck et al. 2017, Liu et al. 2017). We did not find changes for SSTR2 in our series in cases in which pre-surgical treatment with SRL was performed. The in-deep explanation of our finding requires additional in vitro and in vivo experiments for its confirmation, but if it would be so, it would open new potential therapeutic options, as the combined and sequential treatment with first-generation SRL followed by pasireotide may be a new possibility which has never been previously tested.
Our study has some weaknesses, as it has a retrospective design in which not all the patients had all the complete information for full analyses and that not all the markers described have been included in the molecular evaluation, such as the measurement of SST truncated receptors (Luque et al. 2015). On the other hand, the strength of our study is that a relatively large cohort with no special selection bias from real clinical practice was included.
Thus, in summary, we conclude that E-cadherin expression is different between responder, partial responder and resistant patients to first-generation SRL and that it may be considered as a potential biomarker for predictive response to SRL, to be included in therapeutic algorithm in acromegaly patients after surgical failure in order to perform a more personalized and predictive treatment of acromegaly.
Supplementary materials
This is linked to the online version of the paper at https://doi.org/10.1530/ERC-18-0565.
Declaration of interest
M P D, M S, S M W, G S, I S, E V, A P, C P, G O, C V, R C, C F M, I B and M M declare to have received funding from Novartis through the REMAH consortium for research purposes and from Novartis, Ipsen and Pfizer as lecturers. The other authors declare no conflicts of interest.
Funding
This work was supported by the Instituto de Salud Carlos III, co-funded by FEDER (PMP15/00027) and Novartis through the REMAH (Registro Español Molecular de Adenomas Hipofisarios) consortium to the SEEN (Sociedad Española de Endocrinología y Nutrición).
Acknowledgements
REMAH investigators: Magdalena Adrados, Pedro Martínez Flores, Ana María Ramos Leví and Ana Serrano-Somavilla (Hospital Universitario de La Princesa); María Paz de Miguel Novoa (Hospital Clínico San Carlos); Mercedes García Villanueva, Pedro Iglesias and Víctor Rodríguez Berrocal (Hospital Ramón y Cajal); Concepción Blanco (Hospital Príncipe de Asturias); Carmen Alameda Hernando (Hospital Infanta Sofía); Jesús Miguel Pérez Luis (Hospital Universitario de Tenerife); Rogelio García Centeno and Begoña Iza (Hospital Gregorio Marañón); Cristina Alvarez-Escolá, Álvaro Pérez Zamarrón and Carlos Pérez López (Hospital La Paz); José F. Alén, María Calatayud Gutiérrez and Igor Paredes Sansinenea (Hospital 12 de Octubre); Álvaro Otero, José María Recio and Pablo Sousa (Hospital Clínico de Salamanca); Esperanza Aguilló Gutiérrez, Luciano Bances, Fernando L. Calvo Gracia and Fernando Comuñas (Hospital Clínico de Zaragoza); Iris Crespo, Luciana Mª Martel, Pau Fernando Muñoz, Eugenia Resmini, Olga Roig, Alicia Santos, Pere Tresserras, Carlos Asencio, Juan-Ramón Gras and Juan Maria Montserrat (Hospital de Sant Pau); Alberto Torres and Noemí Vidal (Hospital de Bellvitge); Elena Guanyabens and Gemma Sesmilo (Hospital Universitari Quirón Dexeus); Priscilla Cajas and Guillem Cuatrecasas (Centro Médico Teknon); Joaquim Enseñat, Mireia Mora, Felicia Hansu and Irene Halperin (Hospital Clínic Barcelona); Núria Villalmanzo, Laura Cecenarro and Cristina Hostalot (Hospital Germans Trias); Pedro Riesgo (Hospital de La Ribera); Juan Antonio Simal-Julian (Hospital La Fe); Hernán Sandoval and Cristina Lamas (Hospital General de Albacete); Javier Abarca, Nieves Arias Mendoza, Ruth Sánchez Ortiga, Irene Monjas and Ignacio Aranda (Hospital General de Alicante); Teresa Pedro Font (Hospital de Denia); Isabel Alonso Troncoso and Pablo Fernández Catalina (Complejo Hospitalario Pontevedra); Rosa María Álvarez San Martín, Rocío Villar and María D. Ballesteros (Complejo Asistencial de León); Sihara Pérez Romero (Universidad Santiago Compostela); Eva Fernández Rodríguez, Alfredo García-Allut and Ramón Serramito (Hospital Clínico Universitario de Santiago); Alma Prieto (Hospital el Bierzo, León); Laura Cotovad Bellas (Complejo Hospitalario Arquitecto Marcide) and José-Ignacio Vidal Pardo (Complejo Hospitalario Xeral Calde).
References
Al-Brahim NYY & Asa SL 2006 My approach to pathology of the pituitary gland. Journal of Clinical Pathology 59 1245–1253. (https://doi.org/10.1136/jcp.2005.031187)
Bakhtiar Y, Hirano H, Arita K, Yunoue S, Fujio S, Tominaga A, Sakoguchi T, Sugiyama K, Kurisu K, Yasufuku-Takano J, et al. 2010 Relationship between cytokeratin staining patterns and clinico-pathological features in somatotropinomae. European Journal of Endocrinology 163 531–539. (https://doi.org/10.1530/EJE-10-0586)
Casar-Borota O, Heck A, Schulz S, Nesland JM, Ramm-Pettersen J, Lekva T, Alafuzoff I & Bollerslev J 2013 Expression of SSTR2a, but not of SSTRs 1, 3, or 5 in somatotroph adenomas assessed by monoclonal antibodies was reduced by octreotide and correlated with the acute and long-term effects of octreotide. Journal of Clinical Endocrinology and Metabolism 98 E1730–E1739. (https://doi.org/10.1210/jc.2013-2145)
Chahal HS, Trivellin G, Leontiou CA, Alband N, Fowkes RC, Tahir A, Igreja SC, Chapple JP, Jordan S, Lupp A, et al. 2012 Somatostatin analogs modulate AIP in somatotroph adenomas: the role of the ZAC1 pathway. Journal of Clinical Endocrinology and Metabolism 97 E1411–E1420. (https://doi.org/10.1210/jc.2012-1111)
Chanson P & Salenave S 2008 Acromegaly. Orphanet Journal of Rare Diseases 3 17. (https://doi.org/10.1186/1750-1172-3-17)
Coelho MCA, Vasquez ML, Wildemberg LE, Vázquez-Borrego MC, Bitana L, Camacho AHDS, Silva D, Ogino LL, Ventura N, Chimelli L, et al. 2018 Molecular evidence and clinical importance of β-arrestins expression in patients with acromegaly. Journal of Cellular and Molecular Medicine 22 2110–2116. (https://doi.org/10.1111/jcmm.13427)
Efstathiadou ZA, Bargiota A, Chrisoulidou A, Kanakis G, Papanastasiou L, Theodoropoulou A, Tigas SK, Vassiliadi DA, Alevizaki M & Tsagarakis S 2015 Impact of gsp mutations in somatotroph pituitary adenomas on growth hormone response to somatostatin analogs: a meta-analysis. Pituitary 18 861–867. (https://doi.org/10.1007/s11102-015-0662-5)
Fougner SL, Bollerslev J, Latif F, Hald JK, Lund T, Ramm-Pettersen J & Berg JP 2008 Low levels of raf kinase inhibitory protein in growth hormone-secreting pituitary adenomas correlate with poor response to octreotide treatment. Journal of Clinical Endocrinology and Metabolism 93 1211–1216. (https://doi.org/10.1210/jc.2007-2272)
Fougner SL, Lekva T, Borota OC, Hald JK, Bollerslev J & Berg JP 2010 The expression of E-cadherin in somatotroph pituitary adenomas is related to tumor size, invasiveness, and somatostatin analog response. Journal of Clinical Endocrinology and Metabolism 95 2334–2342. (https://doi.org/10.1210/jc.2009-2197)
Fougner SL, Casar-Borota O, Heck A, Berg JP & Bollerslev J 2012 Adenoma granulation pattern correlates with clinical variables and effect of somatostatin analogue treatment in a large series of patients with acromegaly. Clinical Endocrinology 76 96–102. (https://doi.org/10.1111/j.1365-2265.2011.04163.x)
Franck SE, Gatto F, van der Lely AJ, Janssen JAMJL, Dallenga AHG, Nagtegaal AP, Hofland LJ & Neggers SJCMM 2017 Somatostatin receptor expression in GH-secreting pituitary adenomas treated with long-acting somatostatin analogues in combination with pegvisomant. Neuroendocrinology 105 44–53. (https://doi.org/10.1159/000448429)
Freda PU 2002 Somatostatin analogs in acromegaly. Journal of Clinical Endocrinology and Metabolism 87 3013–3018. (https://doi.org/10.1210/jcem.87.7.8665)
Gadelha MR 2015 A paradigm shift in the medical treatment of acromegaly: from a ‘trial and error’ to a personalized therapeutic decision-making process. Clinical Endocrinology 83 1–2. (https://doi.org/10.1111/cen.12797)
Gadelha MR, Kasuki L & Korbonits M 2013 Novel pathway for somatostatin analogs in patients with acromegaly. Trends in Endocrinology and Metabolism 24 238–246. (https://doi.org/10.1016/j.tem.2012.11.007)
Gadelha MR, Wildemberg LE, Bronstein MD, Gatto F & Ferone D 2017 Somatostatin receptor ligands in the treatment of acromegaly. Pituitary 20 100–108. (https://doi.org/10.1007/s11102-017-0791-0)
Gatto F, Feelders RA, van der Pas R, Kros JM, Waaijers M, Sprij-Mooij D, Neggers SJCMM, van der Lelij AJ, Minuto F, Lamberts SWJ, et al. 2013 Immunoreactivity score using an anti-sst2A receptor monoclonal antibody strongly predicts the biochemical response to adjuvant treatment with somatostatin analogs in acromegaly. Journal of Clinical Endocrinology and Metabolism 98 E66–E71. (https://doi.org/10.1210/jc.2012-2609)
Gatto F, Biermasz NR, Feelders RA, Kros JM, Dogan F, van der Lely AJ, Neggers SJCMM, Lamberts SWJ, Pereira AM, Ferone D, et al. 2016 Low beta-arrestin expression correlates with the responsiveness to long-term somatostatin analog treatment in acromegaly. European Journal of Endocrinology 174 651–662. (https://doi.org/10.1530/EJE-15-0391)
Gatto F, Feelders RA, Franck SE, van Koetsveld PM, Dogan F, Kros JM, Neggers SJCMM, van der Lely AJ, Lamberts SWJ, Ferone D, et al. 2017 In vitro head-to-head comparison Between octreotide and pasireotide in GH-secreting pituitary adenomas. Journal of Clinical Endocrinology and Metabolism 102 2009–2018. (https://doi.org/10.1210/jc.2017-00135)
Gonzalez B, Vargas G, Ramirez C, Asa S, Cheng S, Sandoval C & Mercado M 2014 Cytoplasmic expression of SSTR2 and 5 by immunohistochemistry and by RT/PCR is not associated with the pharmacological response to octreotide. Endocrinologia y Nutricion 61 523–530. (https://doi.org/10.1016/j.endonu.2014.05.006)
Ibáñez-Costa A, Gahete MD, Rivero-Cortés E, Rincón-Fernández D, Nelson R, Beltrán M, de la Riva A, Japón MA, Venegas-Moreno E, Gálvez MÁ, et al. 2015 In1-ghrelin splicing variant is overexpressed in pituitary adenomas and increases their aggressive features. Scientific Reports 5 8714. (https://doi.org/10.1038/srep08714)
Kasuki L, Wildemberg LEA, Neto LV, Marcondes J, Takiya CM & Gadelha MR 2013 Ki-67 is a predictor of acromegaly control with octreotide LAR independent of SSTR2 status and relates to cytokeratin pattern. European Journal of Endocrinology 169 217–223. (https://doi.org/10.1530/EJE-13-0349)
Kasuki L, Wildemberg LE & Gadelha MR 2018 MANAGEMENT of ENDOCRINE DISEASE: Personalized medicine in the treatment of acromegaly. European Journal of Endocrinology 178 R89–R100. (https://doi.org/10.1530/EJE-17-1006)
Katznelson L, Laws ER, Melmed S, Molitch ME, Murad MH, Utz A, Wass JAH & Endocrine Society 2014 Acromegaly: an endocrine society clinical practice guideline. Journal of Clinical Endocrinology and Metabolism 99 3933–3951. (https://doi.org/10.1210/jc.2014-2700)
Kiseljak-Vassiliades K, Xu M, Mills TS, Smith EE, Silveira LJ, Lillehei KO, Kerr JM, Kleinschmidt-DeMasters BK & Wierman ME 2015a Differential somatostatin receptor (SSTR) 1-5 expression and downstream effectors in histologic subtypes of growth hormone pituitary tumors. Molecular and Cellular Endocrinology 417 73–83. (https://doi.org/10.1016/j.mce.2015.09.016)
Kiseljak-Vassiliades K, Carlson NE, Borges MT, Kleinschmidt-DeMasters BK, Lillehei KO, Kerr JM & Wierman ME 2015b Growth hormone tumor histological subtypes predict response to surgical and medical therapy. Endocrine 49 231–241. (https://doi.org/10.1007/s12020-014-0383-y)
Lekva T, Berg JP, Fougner SL, Olstad OK, Ueland T & Bollerslev J 2012 Gene expression profiling identifies ESRP1 as a potential regulator of epithelial mesenchymal transition in somatotroph adenomas from a large cohort of patients with acromegaly. Journal of Clinical Endocrinology and Metabolism 97 E1506–E1514. (https://doi.org/10.1210/jc.2012-1760)
Lekva T, Berg JP, Heck A, Lyngvi Fougner S, Olstad OK, Ringstad G, Bollerslev J & Ueland T 2013 Attenuated RORC expression in the presence of EMT progression in somatotroph adenomas following treatment with somatostatin analogs is associated with poor clinical recovery. PLoS ONE 8 e66927. (https://doi.org/10.1371/journal.pone.0066927)
Liu W, Xie L, He M, Shen M, Zhu J, Yang Y, Wang M, Hu J, Ye H, Li Y, et al. 2017 Expression of somatostatin receptor 2 in somatotropinoma correlated with the short-term efficacy of somatostatin analogues. International Journal of Endocrinology 2017 9606985. (https://doi.org/10.1155/2017/9606985)
Luque RM, Ibáñez-Costa A, Neto LV, Taboada GF, Hormaechea-Agulla D, Kasuki L, Venegas-Moreno E, Moreno-Carazo A, Gálvez MÁ, Soto-Moreno A, et al. 2015 Truncated somatostatin receptor variant sst5TMD4 confers aggressive features (proliferation, invasion and reduced octreotide response) to somatotropinomas. Cancer Letters 359 299–306. (https://doi.org/10.1016/j.canlet.2015.01.037)
Mallona I, Díez-Villanueva A, Martín B & Peinado MA 2017 Chainy: an universal tool for standardized relative quantification in real-time PCR. Bioinformatics 33 1411–1413. (https://doi.org/10.1093/bioinformatics/btw839)
Melmed S 2006 Medical progress: acromegaly. New England Journal of Medicine 355 2558–2573. (https://doi.org/10.1056/NEJMra062453)
Melmed S, Bronstein MD, Chanson P, Klibanski A, Casanueva FF, Wass JAH, Strasburger CJ, Luger A, Clemmons DR & Giustina A 2018 A Consensus Statement on acromegaly therapeutic outcomes. Nature Reviews: Endocrinology 14 552–561. (https://doi.org/10.1038/s41574-018-0058-5)
Puig Domingo M 2015 Treatment of acromegaly in the era of personalized and predictive medicine. Clinical Endocrinology 83 3–14. (https://doi.org/10.1111/cen.12731)
Puig-Domingo M, Resmini E, Gomez-Anson B, Nicolau J, Mora M, Palomera E, Martí C, Halperin I & Webb SM 2010 Magnetic resonance imaging as a predictor of response to somatostatin analogs in acromegaly after surgical failure. Journal of Clinical Endocrinology and Metabolism 95 4973–4978. (https://doi.org/10.1210/jc.2010-0573)
Reid TJ, Post KD, Bruce JN, Nabi Kanibir M, Reyes-Vidal CM & Freda PU 2010 Features at diagnosis of 324 patients with acromegaly did not change from 1981 to 2006: acromegaly remains under-recognized and under-diagnosed. Clinical Endocrinology 72 203–208. (https://doi.org/10.1111/j.1365-2265.2009.03626.x)
Rochette C, Castinetti F & Brue T 2016 Acromegaly and Cushing’s disease: persistence of comorbidities after the control of hypersecretion. Annales d’Endocrinologie 77 (Supplement 1) S19–S28. (https://doi.org/10.1016/S0003-4266(17)30074-4)
Rotondo F, Di Ieva A, Kovacs K, Cusimano MD, Syro LV, Diamandis EP & Yousef GM 2015 Human kallikrein 10 in surgically removed human pituitary adenomas. Hormones 14 272–279. (https://doi.org/10.14310/horm.2002.1558)
Taboada GF, Luque RM, Bastos W, Guimarães RFC, Marcondes JB, Chimelli LMC, Fontes R, Mata PJP, Filho PN, Carvalho DP, et al. 2007 Quantitative analysis of somatostatin receptor subtype (SSTR1-5) gene expression levels in somatotropinomas and non-functioning pituitary adenomas. European Journal of Endocrinology 156 65–74. (https://doi.org/10.1530/eje.1.02313)
Vandesompele J, De Preter K, Pattyn F, Poppe B, Van Roy N, De Paepe A & Speleman F 2002 Accurate normalization of real-time quantitative RT-PCR data by geometric averaging of multiple internal control genes. Genome Biology 3 RESEARCH0034. (https://doi.org/10.1186/gb-2002-3-7-research0034)
Venegas-Moreno E, Vazquez-Borrego MC, Dios E, Gros-Herguido N, Flores-Martinez A, Rivero-Cortés E, Madrazo-Atutxa A, Japón MA, Luque RM, Castaño JP, et al. 2018 Association between dopamine and somatostatin receptor expression and pharmacological response to somatostatin analogues in acromegaly. Journal of Cellular and Molecular Medicine 22 1640–1649. (https://doi.org/10.1111/jcmm.13440)
Wildemberg LEA, Vieira Neto L, Costa DF, Nasciutti LE, Takiya CM, Alves LM & Gadelha MR 2012 Validation of immunohistochemistry for somatostatin receptor subtype 2A in human somatotropinomas: comparison between quantitative real time RT-PCR and immunohistochemistry. Journal of Endocrinological Investigation 35 580–584. (https://doi.org/10.3275/7906)
Wildemberg LEA, Neto LV, Costa DF, Nasciuti LE, Takiya CM, Alves LM, Rebora A, Minuto F, Ferone D & Gadelha MR 2013 Low somatostatin receptor subtype 2, but not dopamine receptor subtype 2 expression predicts the lack of biochemical response of somatotropinomas to treatment with somatostatin analogs. Journal of Endocrinological Investigation 36 38–43. (https://doi.org/10.3275/8305)
