ORIGINAL RESEARCH ARTICLE
Metabolomics Biomarkers of Frailty in Elderly Breast Cancer Patients
Cellular Physiology
GIUSEPPE CORONA,1* JERRY POLESEL,2 LUCIA FRATINO,3 GIANMARIA MIOLO,3 FLAVIO RIZZOLIO,1,4 DIANA CRIVELLARI,3 RICCARDO ADDOBBATI,5 SILVIA CERVO,6 GIUSEPPE TOFFOLI1 1
898
Journal of
AND
Experimental and Clinical Pharmacology Division, Department of Translational Research, IRCCS-National Cancer Institute, Aviano, PN, Italy
2
Epidemiology and Biostatistics, IRCCS-National Cancer Institute, Aviano, PN, Italy
3
Medical Oncology Department, IRCCS-National Cancer Institute, Aviano, PN, Italy
4
Sbarro Institute for Cancer Research and Molecular Medicine, Center for Biotechnology, College of Science and Technology, Temple University, Philadelphia, Pennsylvania
5
Metabolic Disease Laboratory, IRCCS-Burlo Garofolo, Children’s Hospital, Trieste, TS, Italy
6
Clinical Pathology Laboratory, IRCCS-National Cancer Institute, Aviano, PN, Italy
Metabolome analysis has emerged as a powerful technique for detecting and define specific physio-pathological phenotypes. In this investigation the diagnostic potential of metabolomics has been applied to better characterize the multiple biochemical alterations that concur in the definition of the frailty phenotype observed in elderly breast cancer patients. The study included 89 women with breast cancer (range 70–97 years) classified as Fit (n ¼ 49), Unfit (n ¼ 23), or Frail (n ¼ 17) according to comprehensive geriatric assessment. The serum metabolomic profile was performed by tandem mass spectrometry and included different classes of metabolites such as amino acids, acylcarnitines, sphingo-, and glycerol-phospolipids. ANOVA was applied to identify the metabolites differing significantly among Fit, Unfit, and Frail patients. In patients carrying the frail phenotype, the amino acid perturbations involve serine, tryptophan, hydroxyproline, histidine, its derivate 3-methyl-hystidine, cystine, and b-aminoisobutyric acid. With regard to lipid metabolism, the frailty phenotype was characterized by a decrease of a wide number of glycerol- and sphingo-phospholipid metabolites. These metabolomics biomarkers may give a further insight into the biochemical processes involved in the development of frailty in breast cancer patients. Moreover, they might be useful to refine the comprehensive geriatric assessment model. J. Cell. Physiol. 229: 898–902, 2014. ß 2013 Wiley Periodicals, Inc.
Cancer treatments and the medical management of older patients are complex and need the close interaction between oncologists and geriatricians in order to provide and optimize treatment options. The physiological reserve of vital organs decrease with age, but the rate of decrease greatly differs from patient to patient. This physiological decline may influence tolerance to cancer chemotherapy, as well as the overall riskbenefit ratio of cancer treatments (Hurria et al., 2011). An objective clinical approach used to capture the functional decline, disability, and comorbidity among the elderly is represented by the comprehensive geriatric assessment (CGA) which classifies elderly patients as Fit, Unfit, and Frail according to essential clinical signs and symptoms (Balducci, 2006). The predictive role of this clinical tool for severe toxicities and mortality has been confirmed in different clinical settings (Hamaker et al., 2011). However, the CGA is essentially a descriptive clinical tool and is therefore unable to measure the molecular and biochemical modifications that underlie the loss of biological functionality. Thus, new and more specific biomarkers to defining the effective physiological conditions of cancer patients are strongly needed. The status of frailty is characterized by biochemical and metabolic impairments involving deregulation of neuromuscular, endocrine, and immune systems that may also be influenced by environmental factors such as physical inactivity, smoking, and an unhealthy diet. Therefore, a ß 2 0 1 3 W I L E Y P E R I O D I C A L S , I N C .
comprehensive investigative approach that takes into account the biological system and their influencing factors is considered more suitable for identifying valuable biomarkers of frailty. The emerging field of metabolomics seems to offer this opportunity
The authors declare no conflict of interest. Contract grant sponsor: Fondazione Umberto Veronesi. Contract grant sponsor: Special Program Molecular Clinical Oncology; Contract grant number: 12214. Contract grant sponsor: European Research Council; Contract grant number: 269051. *Correspondence to: Giuseppe Corona, Clinical and Experimental Pharmacology Division, Department of Translational Research, IRCCS-National Cancer Institute, Via Franco Gallini, 2, 33081 Aviano, PN, Italy. E-mail: [email protected] Manuscript Received: 7 October 2013 Manuscript Accepted: 20 November 2013 Accepted manuscript online in Wiley Online Library (wileyonlinelibrary.com): 27 November 2013. DOI: 10.1002/jcp.24520
METABOLOMICS BIOMARKERS OF FRAILTY
as it is able to assess complex phenotypes that are the net result of genomic, transcriptomic, proteomic, and environmental effects combined with age and other physiopathological conditions (Vinayavekhin et al., 2010). These features make metabolomics a useful tool to better capture the multi-factorial biochemical perturbations occurring in the definition of the frailty phenotype. In this study we explore the potential of a targeted metabolomics approach to better characterize the biochemistry underlining the frail status and to identify new valuable molecular biomarkers of frailty useful to improve the clinical assessment of the elderly patients’ phenotype; this is to be done by analyzing the serum metabolic profile differences among breast cancer patients classified as Fit, Unfit, and Frail according to the CGA. Materials and Methods Patients and study design This study enrolled a total of 95 patients aged 70 years or older with breast cancer who were admitted from 2009 to 2010 to the medical oncology units of our institute however six patients were excluded because of the lack of clinical data, thus leaving 89 patients. Before entering the study, patients signed an informed consent approved by the ethics committee. Patients were evaluated for metabolomics investigation at the moment of diagnosis before surgery or chemotherapy treatments and then classified as Fit, Unfit, and Frail according to the CGA criteria (Extermann and Hurria, 2007). Clinical samples Each patient provided a 10 ml blood sample on the first day of admission. Blood was drawn in the morning between 8 and 10 a.m. after overnight fasting and collected in BD Vacutainer1 tubes coated with silica particles. The sample blood was inverted five times and allowed 30 min clotting time than centrifuged at 2,000g for 10 min 4˚C and serum aliquoted and stored at 80˚C. Targeted metabolomic investigations Two different validated targeted metabolomics analytical platforms based on liquid chromatography tandem mass spectrometry were used. The serum profile of amino acids (n ¼ 45) involved in a wide set of biochemical pathways was performed by using iTRAQ1 (ABsciex, Toronto, Canada). The serum profile for acylcarnitines (n ¼ 40), involved in cellular energetic metabolism and the profile of phospholipids (n ¼ 150) involved in the fatty acid metabolism and cellular signaling, were determined by AbsoluteIDQTM method (Biocrates, Innsbruck, Austria). The complete list of the metabolites is reported in Supplementary online data. Serum amino acids profile For amino acid analysis 50 ml of serum were treated with 10 ml of 10% sulfosalicylic acid to precipitate proteins. After mixing for 30 sec, samples were centrifuged for 5 min at 2,000 rpm (7,000g) and 10 ml of supernatant was diluted with 40 ml 0.45 M borate buffer, pH 8.5. Subsequently, 10 ml of the solution was derivatized with 5 ml of iTRAQ1-115 reagent by incubation at room temperature for 30 min. The reaction was stopped by adding 5 ml of 1.2% hydroxylamine solution. Samples were then dried in Z2 speed vacuum (Genevac, Suffolk, UK) and reconstituted with 32 ml of iTRAQ1 reagent 114-labeled internal standard mix. The isobaric amino acid derivatives were separated by liquid chromatography using C18, 5 m, 4.6 i.d. 150 mm column (ABsciex, Toronto, Canada) running at 50˚C. The mobile phase consisted of 0.1% formic acid and 0.01% heptafluorobutyric acid in water (solvent A) and acetonitrile (solvent B), respectively. The column was equilibrated in JOURNAL OF CELLULAR PHYSIOLOGY
98% solvent A and the gradient was 98–72% A over 10 min, 72–0% A over 0.1 min, hold at 100% B for 5.9 min. A flow rate of 800 ml min1 was used; injection volume was 2 ml. Quantitative analysis was performed by an internal standard method using the signal of multiple reaction-monitoring (MRM) of each amino acid derivatized with 114 and 115 (Internal standard) reagents, respectively, and concentrations were calculated by AnalystTM 1.4.2 software. Serum acylcarnitines and phospholipids profile Plasma acylcarnitines, sphingo-, and glycerol-phospholipids were determined using the analytical protocol of the AbsoluteIDQ1 analytical Kit by flow injection analysis coupled with multiple reaction monitoring scans (FIA-MS/MS). Quantitative determination was performed by MRM, neutral loss, and precursor ion scans in both positive and negative ion mode. The Tuboionspray source was set at the following parameters in accordance with the manufacturer’s indications: Ion spray voltage (IS) 1,500 V; auxiliary gas temperature (TEM) 700˚C; curtain gas (CUR), nebulizer gas (GS1); auxiliary gas (GS2) 20 and 70 arbitrary units, respectively; collision gas medium; entrance potential (EP) 10 V; declustering potential (DP) 20 V; collision energy (CE) 30 V; and collision cell exit potential (CXP) 5 V. Statistical analysis Differences in mean values of metabolites among Fit, Unfit, and Frail were evaluated through the F-test using ANOVA by fixing appropriate linear contrasts to test for increasing/decreasing trends of mean levels. A residual model was applied to assess the differences among metabolite levels taking into account the possible effect of aging. Residuals were first calculated through a linear model with metabolite concentration as a dependent variable and age as an independent variable. Residuals were then analyzed by ANOVA to evaluate age-adjusted trends. Results
Eighty-nine elderly patients with breast cancer (median age 77 years; age range 70–97 years) were clinically stratified according to the CGA criteria in Fit (n ¼ 49); Unfit (n ¼ 23); and Frail (n ¼ 17). The clinical characteristics of patients are shown in Table 1. Both the mean age and the performance status were progressively higher from Fit to Frail (P < 0.01). No statistical differences were reported regarding tumor size and regional lymph node involvement and tumor grading among the three groups. Targeted metabolomics investigations
The metabolites that best differentiated the Frail and Unfit cancer patients from the Fit ones were specific amino acids and acylcarnitines and mainly involved the deregulation of phospholipids. In Figure 1 the percentage of variation of the serum concentration of the more significant metabolites (P 0.05) is reported as observed in Frail and Unfit patients compared to the levels observed in Fit patients used as a control group. The absolute serum concentrations (mean SD) of each metabolite found to be significantly different among Fit, Unfit, and Frail patients are reported in Supplementary online data. Amino acid
The frailty phenotype showed different specific amino acid alterations. The b-aminoisobutyric acid (bAib) in Unfit and Frail patients increased by 6.9% and 61.6%, respectively. The histidine (His) levels decreased almost to the same extent in Unfit (12.8%) and Frail (13.4%), while those of its derivate 3methyl-hystidine (3MHis) increased by 9.0% and 46.7%,
899
900
CORONA ET AL.
TABLE 1. Distribution of 89 female breast cancer patients according to tumor characteristics and comprehensive geriatric assessment
FRAIL (n ¼ 17)
C16:1, C22:1, C24:1 SM(OH) for the sphingolipids class and C32 to C38 derivatives that belong to the group of acyl-ether fatty acids for the PC derivatives (Fig. 1). Alternately, the increase of circulating levels of acylcarnitines observed in Frail patients was only associated to patients age. Discussion
Geriatric assessment FIT (n ¼ 49) n Age
Metabolomics Biomarkers of Frailty in Elderly Breast Cancer Patients
Cellular Physiology
GIUSEPPE CORONA,1* JERRY POLESEL,2 LUCIA FRATINO,3 GIANMARIA MIOLO,3 FLAVIO RIZZOLIO,1,4 DIANA CRIVELLARI,3 RICCARDO ADDOBBATI,5 SILVIA CERVO,6 GIUSEPPE TOFFOLI1 1
898
Journal of
AND
Experimental and Clinical Pharmacology Division, Department of Translational Research, IRCCS-National Cancer Institute, Aviano, PN, Italy
2
Epidemiology and Biostatistics, IRCCS-National Cancer Institute, Aviano, PN, Italy
3
Medical Oncology Department, IRCCS-National Cancer Institute, Aviano, PN, Italy
4
Sbarro Institute for Cancer Research and Molecular Medicine, Center for Biotechnology, College of Science and Technology, Temple University, Philadelphia, Pennsylvania
5
Metabolic Disease Laboratory, IRCCS-Burlo Garofolo, Children’s Hospital, Trieste, TS, Italy
6
Clinical Pathology Laboratory, IRCCS-National Cancer Institute, Aviano, PN, Italy
Metabolome analysis has emerged as a powerful technique for detecting and define specific physio-pathological phenotypes. In this investigation the diagnostic potential of metabolomics has been applied to better characterize the multiple biochemical alterations that concur in the definition of the frailty phenotype observed in elderly breast cancer patients. The study included 89 women with breast cancer (range 70–97 years) classified as Fit (n ¼ 49), Unfit (n ¼ 23), or Frail (n ¼ 17) according to comprehensive geriatric assessment. The serum metabolomic profile was performed by tandem mass spectrometry and included different classes of metabolites such as amino acids, acylcarnitines, sphingo-, and glycerol-phospolipids. ANOVA was applied to identify the metabolites differing significantly among Fit, Unfit, and Frail patients. In patients carrying the frail phenotype, the amino acid perturbations involve serine, tryptophan, hydroxyproline, histidine, its derivate 3-methyl-hystidine, cystine, and b-aminoisobutyric acid. With regard to lipid metabolism, the frailty phenotype was characterized by a decrease of a wide number of glycerol- and sphingo-phospholipid metabolites. These metabolomics biomarkers may give a further insight into the biochemical processes involved in the development of frailty in breast cancer patients. Moreover, they might be useful to refine the comprehensive geriatric assessment model. J. Cell. Physiol. 229: 898–902, 2014. ß 2013 Wiley Periodicals, Inc.
Cancer treatments and the medical management of older patients are complex and need the close interaction between oncologists and geriatricians in order to provide and optimize treatment options. The physiological reserve of vital organs decrease with age, but the rate of decrease greatly differs from patient to patient. This physiological decline may influence tolerance to cancer chemotherapy, as well as the overall riskbenefit ratio of cancer treatments (Hurria et al., 2011). An objective clinical approach used to capture the functional decline, disability, and comorbidity among the elderly is represented by the comprehensive geriatric assessment (CGA) which classifies elderly patients as Fit, Unfit, and Frail according to essential clinical signs and symptoms (Balducci, 2006). The predictive role of this clinical tool for severe toxicities and mortality has been confirmed in different clinical settings (Hamaker et al., 2011). However, the CGA is essentially a descriptive clinical tool and is therefore unable to measure the molecular and biochemical modifications that underlie the loss of biological functionality. Thus, new and more specific biomarkers to defining the effective physiological conditions of cancer patients are strongly needed. The status of frailty is characterized by biochemical and metabolic impairments involving deregulation of neuromuscular, endocrine, and immune systems that may also be influenced by environmental factors such as physical inactivity, smoking, and an unhealthy diet. Therefore, a ß 2 0 1 3 W I L E Y P E R I O D I C A L S , I N C .
comprehensive investigative approach that takes into account the biological system and their influencing factors is considered more suitable for identifying valuable biomarkers of frailty. The emerging field of metabolomics seems to offer this opportunity
The authors declare no conflict of interest. Contract grant sponsor: Fondazione Umberto Veronesi. Contract grant sponsor: Special Program Molecular Clinical Oncology; Contract grant number: 12214. Contract grant sponsor: European Research Council; Contract grant number: 269051. *Correspondence to: Giuseppe Corona, Clinical and Experimental Pharmacology Division, Department of Translational Research, IRCCS-National Cancer Institute, Via Franco Gallini, 2, 33081 Aviano, PN, Italy. E-mail: [email protected] Manuscript Received: 7 October 2013 Manuscript Accepted: 20 November 2013 Accepted manuscript online in Wiley Online Library (wileyonlinelibrary.com): 27 November 2013. DOI: 10.1002/jcp.24520
METABOLOMICS BIOMARKERS OF FRAILTY
as it is able to assess complex phenotypes that are the net result of genomic, transcriptomic, proteomic, and environmental effects combined with age and other physiopathological conditions (Vinayavekhin et al., 2010). These features make metabolomics a useful tool to better capture the multi-factorial biochemical perturbations occurring in the definition of the frailty phenotype. In this study we explore the potential of a targeted metabolomics approach to better characterize the biochemistry underlining the frail status and to identify new valuable molecular biomarkers of frailty useful to improve the clinical assessment of the elderly patients’ phenotype; this is to be done by analyzing the serum metabolic profile differences among breast cancer patients classified as Fit, Unfit, and Frail according to the CGA. Materials and Methods Patients and study design This study enrolled a total of 95 patients aged 70 years or older with breast cancer who were admitted from 2009 to 2010 to the medical oncology units of our institute however six patients were excluded because of the lack of clinical data, thus leaving 89 patients. Before entering the study, patients signed an informed consent approved by the ethics committee. Patients were evaluated for metabolomics investigation at the moment of diagnosis before surgery or chemotherapy treatments and then classified as Fit, Unfit, and Frail according to the CGA criteria (Extermann and Hurria, 2007). Clinical samples Each patient provided a 10 ml blood sample on the first day of admission. Blood was drawn in the morning between 8 and 10 a.m. after overnight fasting and collected in BD Vacutainer1 tubes coated with silica particles. The sample blood was inverted five times and allowed 30 min clotting time than centrifuged at 2,000g for 10 min 4˚C and serum aliquoted and stored at 80˚C. Targeted metabolomic investigations Two different validated targeted metabolomics analytical platforms based on liquid chromatography tandem mass spectrometry were used. The serum profile of amino acids (n ¼ 45) involved in a wide set of biochemical pathways was performed by using iTRAQ1 (ABsciex, Toronto, Canada). The serum profile for acylcarnitines (n ¼ 40), involved in cellular energetic metabolism and the profile of phospholipids (n ¼ 150) involved in the fatty acid metabolism and cellular signaling, were determined by AbsoluteIDQTM method (Biocrates, Innsbruck, Austria). The complete list of the metabolites is reported in Supplementary online data. Serum amino acids profile For amino acid analysis 50 ml of serum were treated with 10 ml of 10% sulfosalicylic acid to precipitate proteins. After mixing for 30 sec, samples were centrifuged for 5 min at 2,000 rpm (7,000g) and 10 ml of supernatant was diluted with 40 ml 0.45 M borate buffer, pH 8.5. Subsequently, 10 ml of the solution was derivatized with 5 ml of iTRAQ1-115 reagent by incubation at room temperature for 30 min. The reaction was stopped by adding 5 ml of 1.2% hydroxylamine solution. Samples were then dried in Z2 speed vacuum (Genevac, Suffolk, UK) and reconstituted with 32 ml of iTRAQ1 reagent 114-labeled internal standard mix. The isobaric amino acid derivatives were separated by liquid chromatography using C18, 5 m, 4.6 i.d. 150 mm column (ABsciex, Toronto, Canada) running at 50˚C. The mobile phase consisted of 0.1% formic acid and 0.01% heptafluorobutyric acid in water (solvent A) and acetonitrile (solvent B), respectively. The column was equilibrated in JOURNAL OF CELLULAR PHYSIOLOGY
98% solvent A and the gradient was 98–72% A over 10 min, 72–0% A over 0.1 min, hold at 100% B for 5.9 min. A flow rate of 800 ml min1 was used; injection volume was 2 ml. Quantitative analysis was performed by an internal standard method using the signal of multiple reaction-monitoring (MRM) of each amino acid derivatized with 114 and 115 (Internal standard) reagents, respectively, and concentrations were calculated by AnalystTM 1.4.2 software. Serum acylcarnitines and phospholipids profile Plasma acylcarnitines, sphingo-, and glycerol-phospholipids were determined using the analytical protocol of the AbsoluteIDQ1 analytical Kit by flow injection analysis coupled with multiple reaction monitoring scans (FIA-MS/MS). Quantitative determination was performed by MRM, neutral loss, and precursor ion scans in both positive and negative ion mode. The Tuboionspray source was set at the following parameters in accordance with the manufacturer’s indications: Ion spray voltage (IS) 1,500 V; auxiliary gas temperature (TEM) 700˚C; curtain gas (CUR), nebulizer gas (GS1); auxiliary gas (GS2) 20 and 70 arbitrary units, respectively; collision gas medium; entrance potential (EP) 10 V; declustering potential (DP) 20 V; collision energy (CE) 30 V; and collision cell exit potential (CXP) 5 V. Statistical analysis Differences in mean values of metabolites among Fit, Unfit, and Frail were evaluated through the F-test using ANOVA by fixing appropriate linear contrasts to test for increasing/decreasing trends of mean levels. A residual model was applied to assess the differences among metabolite levels taking into account the possible effect of aging. Residuals were first calculated through a linear model with metabolite concentration as a dependent variable and age as an independent variable. Residuals were then analyzed by ANOVA to evaluate age-adjusted trends. Results
Eighty-nine elderly patients with breast cancer (median age 77 years; age range 70–97 years) were clinically stratified according to the CGA criteria in Fit (n ¼ 49); Unfit (n ¼ 23); and Frail (n ¼ 17). The clinical characteristics of patients are shown in Table 1. Both the mean age and the performance status were progressively higher from Fit to Frail (P < 0.01). No statistical differences were reported regarding tumor size and regional lymph node involvement and tumor grading among the three groups. Targeted metabolomics investigations
The metabolites that best differentiated the Frail and Unfit cancer patients from the Fit ones were specific amino acids and acylcarnitines and mainly involved the deregulation of phospholipids. In Figure 1 the percentage of variation of the serum concentration of the more significant metabolites (P 0.05) is reported as observed in Frail and Unfit patients compared to the levels observed in Fit patients used as a control group. The absolute serum concentrations (mean SD) of each metabolite found to be significantly different among Fit, Unfit, and Frail patients are reported in Supplementary online data. Amino acid
The frailty phenotype showed different specific amino acid alterations. The b-aminoisobutyric acid (bAib) in Unfit and Frail patients increased by 6.9% and 61.6%, respectively. The histidine (His) levels decreased almost to the same extent in Unfit (12.8%) and Frail (13.4%), while those of its derivate 3methyl-hystidine (3MHis) increased by 9.0% and 46.7%,
899
900
CORONA ET AL.
TABLE 1. Distribution of 89 female breast cancer patients according to tumor characteristics and comprehensive geriatric assessment
FRAIL (n ¼ 17)
C16:1, C22:1, C24:1 SM(OH) for the sphingolipids class and C32 to C38 derivatives that belong to the group of acyl-ether fatty acids for the PC derivatives (Fig. 1). Alternately, the increase of circulating levels of acylcarnitines observed in Frail patients was only associated to patients age. Discussion
Geriatric assessment FIT (n ¼ 49) n Age
