Proteomic Serum Profile of Fatigued Men Receiving Localized External Beam Radiation Therapy for Non-Metastatic Prostate Cancer

Abstract

Context

Fatigue is the most distressing side effect of radiation therapy (RT) and its progression etiology is unknown.

Objectives

This study describes proteome changes from sera of fatigued men with non-metastatic prostate cancer receiving external beam radiation therapy (EBRT).

Methods

Fatigue scores, measured by the Functional Assessment of Chronic Illness Therapy-Fatigue, and serum were collected from 12 subjects at baseline (prior to EBRT) and at midpoint (day21) of EBRT. Depleted sera from both time points were analyzed using two-dimensional difference gel electrophoresis, and up/down regulated proteins were identified using liquid chromatography-tandem mass spectrometry. Western blot analyses confirmed the protein changes observed.

Results

Results showed that apolipoprotein (Apo)A1, ApoE, and transthyretin (TTR) consistently changed from baseline (day 0) to midpoint (day 21). The mean ApoE level of subjects with high change in fatigue (HF: n=9) increased significantly from baseline to midpoint and were higher than in subjects with no change in fatigue. The mean ApoA1 level was higher in HF subjects at baseline and at midpoint than in NF subjects at both time points. The mean TTR level of NF subjects was higher at baseline and midpoint than in HF subjects.

Conclusion

These ApoE, ApoA1, and TTR results may assist in understanding pathways that can explain fatigue progression etiology in this clinical population.

Introduction

Prostate cancer is the most common type of cancer and second leading cause of death among men in the U.S.¹ External beam radiation therapy (EBRT) is one of the treatment modalities preferred by patients with non-metastatic prostate cancer.² Up to 71% of prostate cancer patients complain of fatigue during EBRT, contributing to the decline of their health-related quality of life.^3,4 Fatigue related to cancer and its treatment, also known as cancer-related fatigue (CRF), is identified as the most distressing symptom reported by patients.⁵ CRF is defined as a subjective feeling of overwhelming exhaustion that is not relieved by rest, interfering with the performance of a person’s daily activities.⁶ Evidence suggests that CRF can become a chronic condition post cancer treatment, even impacting the quality of life of disease-free cancer survivors.⁷ CRF remains poorly managed and its etiology remains elusive.

Proteomic-based techniques have been used to identify several biomarkers to diagnose many types of cancer such as early stage ovarian cancer,⁸ prostate cancer,⁹ gastric cancer,¹⁰ and colorectal cancer.¹¹ These methods also help in identifying proteins that are involved in the underlying mechanisms of symptoms such as neuropathic pain,¹² depression,^13,14 and chronic fatigue.¹⁵ Recently, one study used the surface-enhanced laser desorption/ionization (SELDI) technique, coupled with the use of one-dimensional gels and a trypsin digestion method using liquid chromatography, a proteomic methodology to identify possible biomarkers for CRF.¹⁶ To our knowledge, no study has used the two-dimensional difference gel electrophoresis (2-D DIGE) technique to describe the changes in serum proteome that accompanies the clinically significant change in fatigue symptoms experienced by men with non-metastatic prostate cancer while receiving EBRT. This is an unbiased, hypothesis-generating approach, which can potentially identify proteomic markers that can explain the physiologic mechanisms behind the development or worsening of fatigue symptom in this clinical population.

Methods

Sample

Subjects with non-metastatic prostate cancer enrolled in an actively recruiting, National Institutes of Health (NIH), Institutional Review Board-approved protocol (NCT00852111) were included in this analysis. Subjects were 18 years or older, diagnosed with non-metastatic prostate cancer with or without a history of prostatectomy, and scheduled to receive EBRT with or without concurrent androgen deprivation therapy (ADT). Subjects with progressive disease causing significant fatigue; with psychiatric disease within five years; uncorrected hypothyroidism and anemia; taking sedatives, steroids, and nonsteroidal anti-inflammatory agents; and with second malignancies were excluded. The study was conducted at the Magnuson Clinical Research Center, NIH, Bethesda, MD, from May 2009 to September 2011. All participants signed written informed consent prior to participating in the study.

All study participants completed a fatigue questionnaire and had blood drawn at two time points: baseline (prior to EBRT) and at midpoint (day 21 after EBRT initiation). Fatigue level was measured by the Functional Assessment of Chronic Illness Therapy-Fatigue (FACIT-F), a 13-item fatigue-specific subscale. The total score for the FACIT-F ranges from 0 to 52. A higher FACIT-F score indicates less fatigue. This is a validated fatigue measure, which showed good stability (test-retest r = 0.87) and good internal consistency reliability with a coefficient alpha in the mid 90s.¹⁷ To optimize the phenotypic characterization of the study participants, subjects were grouped according to the change in fatigue scores during EBRT: the high fatigue (HF) group were subjects with increasing fatigue symptoms (declining FACIT-F scores) from baseline to midpoint of EBRT; and the no fatigue (NF) group were subjects with no change or with increasing FACIT-F scores between the two time points. Peripheral blood was collected using a serum separator tube (Becton, Dickinson and Company, Franklin Lakes, NJ) at both study time points. The sera were separated, aliquoted into separate Eppendorf tubes, and stored at −80°C until ready for analysis.

Sample Preparation for the 2-D DIGE

Serum samples were processed using the Agilent Multiple Affinity Removal Spin Cartridge (Agilent Technologies, New Castle, DE) to selectively remove the top six abundant proteins. The low abundance protein fraction was concentrated and followed by chloroform/methanol precipitation to pellet the proteins.¹⁸ A total of 50 μg of serum proteins were labeled with one of the three CyDye (Cy2, Cy3, and Cy5) DIGE Fluors (GE Healthcare, Piscataway, NJ) for internal standard (Cy2), control (baseline serum, Cy3) and experimental (midpoint serum, Cy5). The labeled protein samples were combined and focused on 24cm, pH 3–10, and NL IPG strips (GE Healthcare). IPG strips were then loaded onto (12%) SDS-PAGE acrylamide gradient gels (Jule Inc., Milford, CT), sealed with 0.5% agarose solution, and run using an Ettan DALT II System (GE Healthcare). The gels were scanned using a Typhoon 9400 scanner (GE Healthcare) at 100 μm resolution and stained with Sypro ruby dye (Molecular Probes, Eugene, OR).

Relative protein quantification across the control and experimental samples was performed using Progenesis SameSpots software (NonLinear Dynamics, Newcastle, U.K.). Student’s t-tests were used to calculate significant differences in relative abundances of protein spot across several gels. Protein spots that were significantly increased or decreased in relative abundance (greater than 1.5-fold) were chosen for further analysis. Ettan Spot Handling Workstation (GE Healthcare Bio-sciences AB, Uppsala, Sweden) was used to perform automated spot picking followed by in-gel tryptic digestion.

Liquid Chromatography Tandem Mass Spectrometry Analysis and Database Search

Liquid chromatography tandem mass spectrometry (LC-MS/MS) analysis was done using an Eksigent nanoLC-Ultra 1D plus system coupled with LTQ Orbitrap XL mass spectrometer (Thermo Fisher Scientific, San Jose, CA) using CID fragmentation. A protein database search using raw MS/MS data was performed using the Proteome Discoverer v1.3 software (Thermo Fisher Scientific) with MASCOT (Matrix Science, Inc., Boston, MA) as the search engine. The following search criteria were used: Swiss-Prot database; homo sapiens; trypsin; 2 missed cleavages; oxidation (M), deamidation (NQ) as variable; and carbamidomethyl as fixed modification.

Western Blot Analysis

Non-depleted serum samples from all subjects were used for confirmatory Western blot (WB) analysis with primary antibodies against apolipoprotein E (ApoE) (ab1906) (diluted 1:750), transthyretin (TTR) (ab53422) (diluted 1:2000) and apolipoprotein A-I (ApoA1) (ab52945) (diluted 1:500) (Abcam, Inc., Cambridge, MAs) and secondary antibody of IRDye 800 CW infrared Dye (diluted 1:5000) (Li-Cor, Lincoln, NE).

Statistical Analyses

Descriptive statistics were calculated for the participants’ demographic characteristics. Student t-tests using SPSS v. 11.0 software (SPSS Inc., Chicago, IL) were conducted to determine mean differences of fatigue, protein expression, and clinical/demographic characteristics between the groups. A P < 0.05 was considered significant at a 95% confidence level.

Results

Sample Characteristics

Twelve participants were categorized into groups; three subjects in the NF group (mean FACIT-F score: baseline = 44.0±8.0, midpoint = 46.3±9.0) and nine subjects in the HF group (mean FACIT-F score: baseline = 47.0±2.9, midpoint = 36.8±5.5). The mean ages of subjects were very similar (HF = 68.9±5.1 years, NF = 63.0±7.2 years, P =0.14). Half of the subjects had low risk disease with a prostate cancer clinical stage of T1c, and a Gleason score between 6 and 7. At baseline, all participants had a score of 90 on the Karnofsky Performance Status Scale, indicating that they were able to carry out normal activities with minor signs or symptoms of disease. All participants received a total EBRT dose of 7560 Gray. All except one patient received androgen deprivation therapy (ADT) two months before EBRT and none of them received radical prostatectomy before receiving EBRT. No differences in hematocrit levels and depression scores were noted between the fatigue groups at baseline and at midpoint of EBRT. The demographic characteristics of the 12 study participants are shown in Table 1.

Table 1.

Demographic and Clinical Characteristics of Study Sample

Variables	N=12		t-test P-value
	No Fatigue (NF) (n=3)	High Fatigue (HF) (n=9)
Mean age, yrs	63.0 ± 7.2	68.9 ± 5.1	0.14
T-stage
T1c	1	3
T2, NOS	-	1
T2a	2	3
T3a	-	1
Gleason Score
6–7	2	4
8–10	1	5
Karnofsky Performance Status	90.0 ± 0	90.0 ± 0
Testosterone (ng/dL)	272.0 ± 9.9	191.1 ± 191.1	0.40
Thyroid Stimulating Hormone (μIU/mL)	0.9 ± 1.0	2.3 ± 1.3	0.26
PSA level before EBRT	4.3 ± 5.8	5.1 ± 5.2	0.82
Albumin (g/dL)	4.2 ± 0.1	4.1 ± 0.2	0.58
Red Blood cell Count
Day0	4.5 ± 0.4	4.1 ± 0.2	0.08
Day21	4.5 ± 0.4	3.9 ± 0.4	0.06
Hemoglobin (mg/dl)
Day0	14.1 ± 1.4	13.1 ± 0.5	0.33
Day21	13.1 ± 1.2	12.1 ±1.0	0.18
Hematocrit (%)
Day0	41.0 ± 3.2	38.0 ± 1.0	0.24
Day21	37.6 ± 2.7	35.3 ± 2.7	0.25
Depression (Hamilton Rating Scale of Depression)
Day0	0.0 ± 0	0.8 ± 1.0	0.21
Day21	0.0 ± 0.0	2.6 ± 4.0	0.31
Fatigue score (FACIT-F)
Day0	44.3 ± 8.0	47.0 ± 2.9	0.39
Day21	46.3 ± 9.0	36.8 ± 5.5	0.05a

ng/dL = nanogram per deciliter; μIU/mL = micro International Units per milliliter; PSA = prostate specific antigen; g = gram; mg = milligram; Day0 = baseline, before external beam radiation therapy (EBRT), Day21 = midpoint of EBRT; FACIT-F = Functional Assessment of Chronic Illness Therapy-Fatigue.

^aP < 0.05.

2-D DIGE

This unbiased and hypothesis-generating investigation started by running, depleted sera collected from baseline (50 μg) and midpoint (50 μg) of EBRT of one, 59-year-old, Caucasian, HF male in the same 2-D DIGE using a pI of 3–10 for the first dimension and 8–16% gel for the second dimension. This subject was selected based on having the highest change in fatigue (FACIT-F) scores from baseline to midpoint of EBRT. Approximately 200 spots exhibited significant differential protein expression (fold change ≥ 1.5, P <0.05) between the two time points. The sypro ruby stained DIGE image of this gel is provided in Fig. 1. The 200 spots were chosen for LC-MS/MS analysis, identifying proteins in 112 spots (P <0.01) with one or more unique peptides. The rest of the spots did not give any reliable protein identifications because of the low abundance of proteins in those spots. Appendix 1 (available on jpsmjournal.com) contains the list of identified proteins in 112 spots from the chosen gel.

Figure 1.

Sypro ruby stained gel image.

The Sypro ruby stained gel image illustrates numbered positions of peptides, which were identified by LC-MS/MS to detect proteins that consistently changed in the 2 DIGE image of depleted sera collected from baseline to midpoint of external beam radiation therapy (EBRT). These numbers were used to identify the proteins that were detected in the gel spot.

In order to verify the proteomic changes from this one HF patient, the 2-D DIGE procedure was repeated using sera collected from the same time points and pooled sera from both time points used as control from three additional HF Caucasian, male subjects (mean age = 66.25±5.9). Three DIGE spots were found to consistently change from the sera of these four subjects (Fig. 2). These three spots were referred to the LC-MS/MS data and were identified to reflect the overexpression of ApoE (fold change = 1.8, P <0.001) and TTR (fold change = 1.6, P <0.001), and the down regulation of ApoA1 (fold change = 1.5, P <0.001) from baseline to midpoint of EBRT.

Figure 2.

Protein identification of gel spots.

These proteins were found to consistently change in the 2 DIGE image of the sera of 4 HF subjects collected at baseline and at midpoint of external beam radiation therapy.

Confirmatory Test

To confirm the differential expression of these three proteins (ApoE, TTR, ApoA1), WB analyses of non-depleted sera collected from baseline and midpoint of EBRT from 12 subjects (the original four HF subjects used in the DIGE analyses and an additional five HF and three NF subjects) were conducted. The WB band intensities were normalized using the total amount of sera from the loading gel (Fig. 3). Mean values of the normalized band intensities of these three proteins are listed in Table 2. Mean ApoE levels in both groups increased during EBRT, but the mean ApoE level of HF subjects changed significantly from baseline (2.07±1.44) to midpoint of EBRT (7.10±6.91; P =0.03) and were higher than in NF subjects (n=3) at both time points (NF: baseline =1.72±1.41; midpoint = 4.33±3.84). Mean ApoA1 levels increased from baseline to midpoint of EBRT in the two groups; however, the mean Apo1 level was higher in HF subjects at baseline (4.81±6.51) and more significantly at midpoint of EBRT (7.71±6.04, P =0.03) than in NF subjects (baseline = 0.64±0.44; midpoint =1.83±1.74). Mean TTR levels increased from baseline to midpoint of EBRT in HF subjects, but an opposite trend was observed in the mean TTR levels of NF subjects. Mean TTR level was higher in NF subjects at baseline (74.0±28.76, P =0.08) and at midpoint of EBRT (63.4±55.71) than in HF subjects (baseline = 23.6±40.53, midpoint = 38.2±59.64).

Figure 3.

Western Blot band intensities.

Non-normalized Western blot band intensities confirming the three proteins observed to consistently change from the 2 DIGE image of the sera of 12 subjects collected at baseline (Day 0) and at midpoint (Day 21) of external beam radiation therapy.

Table 2.

Mean of Normalized Western Blot Band Intensities

Mean Band Intensity	N=12		t-test P-Values NF vs. HF
	No Fatigue (NF) (n=3)	High Fatigue (HF) (n=9)
ApoE
Day 0	1.72 ± 1.41	2.07 ± 1.44	0.72
Day 21	4.33 ± 3.84	7.10 ± 6.91	0.53
P-value (Day 0 vs. Day 21)	0.21	0.03 a
ApoA1
Day 0	0.64 ± 0.44	4.81 ± 6.51	0.31
Day 21	1.83 ± 1.74	7.71 ± 6.04	0.03 a
P-value (Day 0 vs. Day 21)	0.26	0.16
TTR
Day 0	74.00 ± 28.76	23.55 ± 40.53	0.08
Day 21	63.37 ± 55.71	38.20 ± 59.64	0.54
P-value (Day 0 vs. Day 21)	0.78	0.10

Mean normalized band intensities of apolipoprotein E (ApoE), apolipoprotein A1 (ApoA1) and transthyretin (TTR) at baseline (Day 0) and at midpoint (Day 21) of External Beam Radiation Therapy.

^aP < 0.05.

Discussion

To our knowledge, this is the first longitudinal study to use the 2-D DIGE proteomics technology to demonstrate concomitant changes in expression of novel proteins with changes in fatigue symptoms of prostate cancer patients while receiving localized radiation therapy for non-metastatic prostate cancer. A recent study used a SELDI technique, coupled with the use of one-dimensional gels and a trypsin digestion method using liquid chromatography to identify proteins associated with fatigue experienced by breast cancer survivors.¹⁶ Although the SELDI technique has been used extensively, especially in the early detection of cancer involving low molecular weight cancer-associated proteins,^19,20 it is criticized for a number of reasons including its inability to provide peptide/protein identification,²¹ difficulty in reproducing because of the possibility of over-fitting, and questionable applicability to be used as a routine diagnostic technique.^22–24 The 2-D electrophoresis is a widely used technique to study the proteome.²⁵ This study adapted several approaches to address the inherent challenges in using 2-D DIGE including the application of fractionation methods and affinity electrophoresis to detect proteins with smaller molecular masses and smaller pI values.

This study showed an increase in ApoE expression in all subjects from baseline to midpoint of EBRT. Apolipoproteins play a role in cholesterol transport and have been associated with several important physiological processes necessary to maintain immunoregulation and cognition.²⁶ They also have been found to be involved in proteolytic breakdown of beta-amyloid in Alzheimer’s disease.²⁷ We recently observed a significant association between changes in α-synuclein gene expression from peripheral blood and changes in fatigue symptoms of patients receiving EBRT for non-metastatic prostate cancer.²⁸ A previous animal study has established a direct connection between α-synuclein and ApoE expressions, where upregulation of both proteins caused neurodegeneration in transgenic mice.²⁹ The upregulation of ApoE and α-synuclein were observed in the peripheral blood of this clinical population, and the direct link between these proteins further supports our assertion that neuroprotective mechanisms may explain the etiology behind changes in fatigue intensity during radiation therapy.

ApoA1 levels also were observed to be higher in HF subjects than in NF subjects at baseline but more significantly at midpoint of EBRT. An animal study showed that overexpression of ApoA1 prevented neuroinflammation by reducing glial activation in the brain and also decreased Aβ-induced proinflammatory cytokine/chemokine production in the hippocampus, avoiding learning and memory deficits.³⁰ It also is observed to be highly susceptible to oxidation, which is recently believed to explain the etiology of “chemobrain,” a symptom triggered by the production of reactive oxygen species produced by the oxidation of ApoA1 during chemotherapy, leading to the elevation of tumor necrosis factor (TNF), which crosses the blood-brain barrier contributing to neuronal death.³¹ These observations suggest that increasing ApoA1 protein levels during EBRT in this clinical population is a protective mechanism to counter an acute stressor to prevent neuroinflammation and decrease proinflammatory cytokine production, in which both mechanisms were suggested to potentially explain the etiology behind cancer-related fatigue.^28,32

TTR or pre-albumin is the third differentially expressed protein, whose level is observed in this study to be lower in HF subjects than in NF subjects. This is a negative acute phase protein, where its concentration is reduced during acute phase response.³³ During acute stress such as an acute viral infection, acute phase response is manifested clinically by fatigue, fever, sleep disturbance, mood and cognitive problems, where it is triggered peripherally by an activated immune system mediated by the actions of pro-inflammatory cytokines and the acute phase proteins.³⁴ As a negative acute phase protein, the level of TTR is expected to decrease during acute stress. In this study, TTR levels were lower in HF groups at baseline and at midpoint of EBRT compared with NF subjects, suggesting that an acute phase response manifested by fatigue may be occurring even prior to EBRT. Further investigation is necessary to understand the causes of this observation. The differential expression of these three proteins (ApoE, ApoA1, TTR) is linked with two mechanisms that are known to be associated with cancer-related fatigue: neuroinflammation and proinflammatory cytokine production.

Although the study findings do not directly support the concept that ApoE, ApoA1, or TTR cause the development of fatigue associated with cancer or its treatment, they stimulate important questions for future investigations to understand the pathways behind fatigue development. The current study is limited by having only explored the association between the three proteins and fatigue, and the number of subjects for the fatigue groups is small. Further investigation is necessary to confirm the study findings using a larger, homogeneous sample, and to explore the role of other networks involved in fatigue development.

Pro-inflammatory cytokines such as interleukin (IL)-1, IL-6, and TNF-α have been reported to be associated with CRF.³⁵ Unfortunately, these cytokines are not detected because their molecular mass (typically 10 to 40 kDa) runs below the detection threshold using the DIGE technique and their low abundance makes them almost impossible to detect with mass spectrometry.³⁶ In addition, cytokines bind tightly to albumin;³⁷ hence, cytokines attached to albumin may be removed during the albumin depletion process of the serum samples.³⁸ Detection of proinflammatory markers such as cytokines is best carried out by immunoproteomics profiling such as immunoblotting.³⁹

Conclusion

There is currently no optimal management for fatigue during cancer therapy because of the lack of understanding of the etiology behind its development and, consequently, the inability to identify molecular targets that can serve as interventional options to manage it. Proteomic technology can provide pertinent information in understanding potential molecular pathways behind fatigue development during cancer therapy using an unbiased approach. Replication of these findings in populations with other types of cancer and those receiving other types of cancer treatments is important to confirm the list of differentially expressed proteins reported in this study and to verify their role in fatigue development.