Clinica Chimica Acta 438 (2014) 171–177

Contents lists available at ScienceDirect

Clinica Chimica Acta journal homepage: www.elsevier.com/locate/clinchim

Human epididymis protein 4: Factors of variation Simona Ferraro ⁎, Domitilla Schiumarini, Mauro Panteghini Cattedra di Biochimica Clinica e Biologia Molecolare Clinica, Dipartimento di Scienze Biomediche e Cliniche ‘Luigi Sacco’, Università degli Studi, Milan, Italy

a r t i c l e

i n f o

Article history: Received 13 June 2014 Received in revised form 13 August 2014 Accepted 19 August 2014 Available online 27 August 2014 Keywords: Ovarian cancer Biomarker Biological factors Biological variability

a b s t r a c t Background: Amongst the newly proposed biomarkers for ovarian cancer, serum human epididymis protein 4 (HE4) shows the greatest potential for clinical use. However, systematic appraisals of its biological characteristics are not available. This study sought to critically revise the available literature on biological and lifestyle factors affecting HE4 concentrations in serum to understand their possible influence on the marker interpretation. Methods: A literature search was undertaken on electronic databases and references from retrieved articles. Article results were analyzed by evaluating study design, sample size, statistical approach, employed assay and, when available, by collecting similar information for carbohydrate antigen 125 (CA-125). Results: Several factors may influence serum HE4 concentrations. In contrast to CA-125, higher HE4 concentrations are reported in the elderly. Although no variations in HE4 concentrations can be clearly associated to menopausal status, a strong difference in biomarker biological intra-individual variation according to the fertility status is reported. Smoking and renal function can also significantly influence HE4 results. Conclusion: The knowledge of factors influencing HE4 concentrations is relevant to promote more adequate interpretative criteria for use of this biomarker in the clinical setting. © 2014 Published by Elsevier B.V.

Contents 1. 2. 3.

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Research methodology and identification of studies . . . . . . . . . . . . . Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.1. Factors influencing HE4 concentrations in serum . . . . . . . . . . . 3.1.1. Study design and a priori statistical modelling . . . . . . . . 3.1.2. Gender, age, menopausal status, menstrual cycle and pregnancy 3.1.3. Other factors . . . . . . . . . . . . . . . . . . . . . . . 3.2. Factors influencing CA-125 concentrations in serum . . . . . . . . . 3.3. Wiping out interfering factors to estimate HE4 biological variability . . 4. Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

. . . . . . . . . . .

171 172 172 172 172 172 174 174 176 176 177

1. Introduction

Abbreviations: CA-125, carbohydrate antigen 125; OC, ovarian cancer; HE4, human epididymis protein 4; PPV, positive predictive value; CI, confidence interval; BMI, body mass index; MeSh, medical subject headings; RI, reference interval; GFR, glomerular filtration rate; IVF, in vitro fertilization; LH, luteinizing hormone; FSH, follicle-stimulating hormone; BV, biological variability; RCV, reference change value; NICE, National Institute for Health and Clinical Excellence; PSA, prostate-specific antigen. ⁎ Corresponding author at: Laboratorio Analisi Chimico-Cliniche, Ospedale ‘Luigi Sacco’, Via G.B. Grassi 74, Milano, Italy. Tel.: +39 02 3904 2766; fax: +39 02 3904 2896. E-mail address: [email protected] (S. Ferraro).

http://dx.doi.org/10.1016/j.cca.2014.08.020 0009-8981/© 2014 Published by Elsevier B.V.

Several biological and behavioural factors have been recognized to influence concentrations of circulating tumour markers. For instance, elevations of carbohydrate antigen 125 (CA-125), the classical biomarker for ovarian cancer (OC), over the established diagnostic threshold are associated with individual features (age, menopausal status, ethnicity) and lifestyle factors (smoking, caffeine consumption, use of oral contraceptives, talc use) [1,2]. At least in some cases, these factors at baseline can unpredictability jeopardize the clinical performance of biomarkers, thus limiting their discriminatory capability [3].

172

S. Ferraro et al. / Clinica Chimica Acta 438 (2015) 171–177

Recently, human epididymis protein 4 (HE4) has been introduced as novel tumour marker for OC diagnosis/monitoring and it has been proposed to possibly replace CA-125, due to the far higher specificity and rule in capability [4]. The gene encoding for HE4, WFDC2, is composed by six exons. Five HE4 isoforms have been identified and recognized to define specific patterns differently expressed in neoplasm and normal tissues. Under physiologic conditions, HE4 is reported as protease inhibitor playing a crucial role in sperm maturation [5]. In pathophysiology, HE4 is likely involved in cancer progression and metastases. In particular, HE4 is overexpressed in serous and endometrioid OC [6]. Noteworthy, several authors and assay manufacturers have recommended the interpretation of HE4 by resorting to different decisional levels for women in pre- and post-menopausal status, respectively [7]. Other authors recommended the use of diagnostic algorithms adjusting HE4 (and CA-125) concentrations for the effect of menopausal status [8]. Actually, the estimation of the effect on average biomarker concentrations by each biological factor may become relevant to calibrate the diagnostic algorithms that include biomarker measurements to drive clinical decisions or screening programs [9]. Notably, in the OC framework, several authors have speculated on the clinical and biological inconsistency of fixed threshold levels, and have therefore suggested consideration of the effects of age, ethnicity and other epidemiologic characteristics to generate longitudinal algorithms including OC biomarkers for screening and diagnosis [10]. In the clinical setting the required diagnostic specificity for OC biomarkers should at least overcome the threshold of 95% to limit further expensive confirmatory tests and unnecessary treatments. In particular, by setting a cost-effective positive predictive value (PPV) at 10% (i.e., 10 operations for each case of OC detected) and considering the actual OC prevalence in this framework (~0.04%), the estimated desirable specificity would be 99.6% [4]. However, this goal is quite unattainable for biomarkers, since the specificity of CA-125 and HE4 from metaanalyzed data amounts to 78% (95% confidence interval (CI): 76–80) and 93% (CI: 92–94), respectively [11]. Although the desirable specificity currently represents a far-reaching goal, the rate of false positive results may be reasonably lowered if the absolute marker concentrations are adjusted for the biological and behavioural features affecting them [12]. On the other hands, the recommended approaches for HE4 interpretation, adjusting marker concentrations for age and menopause, entered early into the clinical practice without being definitively supported by robust investigations on the HE4 biological behaviour. Erroneously, the same biological characteristics influencing CA-125 concentrations have been assumed as affecting HE4 values. Last but not least, teasing out the effect of each individual characteristic and of potential biological interactions between two or more factors involves application of complex statistical models. The sample size of the case series is also crucial: it should be as larger as higher is the number of the investigated biological features. Theoretically, the background knowledge of what factors potentially influence a biomarker should be known before the introduction into the clinical setting. Thus, considering HE4 as the most promising novel biomarker for OC and as the main candidate to replace CA-125 in the diagnostic algorithms, we sought to review and critically appraise the available evidence on biological and lifestyle factors affecting HE4 concentrations in serum. For those studies extending the discussion to CA-125, we could cross-validate results on HE4 by using the information already known for CA-125. 2. Research methodology and identification of studies The aim of the search was to identify those articles investigating the effect of individual characteristics [e.g., gender, ethnicity, age, age at menarche, phase of menstrual cycle, pregnancy, menopausal status, body mass index (BMI), smoking habit, renal function, caffeine consumption, use of oral contraceptives] on HE4 serum concentrations under apparently healthy conditions. The peer-reviewed literature

published up to April 30, 2014 was searched using the Medline database (PubMed), with MeSh (Medical Subject Headings) terms “HE4 and biological factor” and with limits “Title/Abstract, English”. Amongst 148 results, we selected 8 pertinent articles [1,2,12–17] providing data on: 1) the association between individual characteristics and HE4 concentrations in serum; 2) the effect of individual characteristics on HE4 concentrations in serum. In addition, we performed further searches, with the same limits, by using instead of “biological factor” the following MeSh terms: “age/ menopausal/menstrual cycle/gender/smoking/creatinine/caffeine/ contraceptive/pregnancy” and from the retrieved papers we selected four additional articles [18–21]. A total of 12 articles were therefore identified. Within references of these articles, 8 studies concerning CA-125 were also selected [9,21–27]. 3. Results 3.1. Factors influencing HE4 concentrations in serum The main features of studies evaluating factors influencing HE4 concentrations are reported in Table 1. In particular, for each study the characteristics of the investigated population (type, age, sample size, setting), data modelling with related results and company/platform of HE4 assays are detailed. 3.1.1. Study design and a priori statistical modelling Most investigations on factors influencing HE4 concentrations were performed by considering apparently healthy individuals recruited from various settings (healthcare employees [13,14,19], screening trials [1,2,12,15,21] or biobank samples [18]), whereas other authors considered hospitalized or out-patients [16,17,20]. In two studies [1,18] univariate analysis was first used for evaluating the association between HE4 concentrations and single factors, whereas multivariate regression models allowed assessing the effect of each covariate on marker concentrations by adjusting for the remaining factors in five studies [1,2,12,18,19]. Seven studies applied a less robust statistical approach by simply evaluating pair-wise comparisons between groups [13–17,20,21]. 3.1.2. Gender, age, menopausal status, menstrual cycle and pregnancy Males have serum HE4 concentrations 7–9% lower than females, but with a more pronounced age-related increase [17,18]. By estimating HE4 reference interval (RI) on 1515 healthy Chinese females, Yang et al. found a statistically significant correlation (P b 0.05) between HE4 concentrations and age, with an increase over 60 years [21]. In a similar investigation on 2182 healthy and mostly young (75% b39 years) women, Park et al. [13] reported a slight increase of HE concentrations with age, whilst Urban et al. [12] showed a statistically significant HE4 increase only in women N 55 years old. Bolstad et al. [18] reported a mild but statistically significant relationship between age and marker concentrations, but with a non-linear contribution. By assuming 20 years of age as baseline, the multivariate model evidenced that HE4 concentrations rise 2% at 30 years, 9% at 40 years, 20% at 50 years, 37% at 60 years, 63% at 70 years and 101% at 80 years of age. In two small studies performed on relatively young women, both Hallamaa et al. [16] and Braga et al. [19] did not find any correlation with age, indirectly confirming that only women aged N 55 years display significantly higher HE4 concentrations. This was also confirmed by Urban et al. [2] who showed in a population with a mean (± SD) age of 65.6 (± 5.2) that there was a strong and positive relationship between HE4 with increasing age [2]. It was unclear if menopausal status per se is able to influence serum HE4 concentrations. Once again, by partitioning the post-menopausal

Table 1 Main features of retrieved studies related to biological factors and other characteristics influencing human epididymis protein 4 (HE4) concentrations in serum. Authors

Population

Age (yr)a

Sample size

Setting of subjects

Data entering

Statistical model

Investigated factorsb

HE4 assay

Lowe et al. [1]

Healthy Post-M at risk of ovarian cancer

55 32–83)

155

Trial

Log conversion

Univariate analysis/ multiple regression

Luminex Multiplexed ELISA

Anastasi et al. [14]

Healthy Pre-M

40

University employees

Original data

t-Test

Escudero et al. [17]

Healthy individuals

31 (20–49) 52 (20–91)

Age, age at menarche, height, weight, BMI, ethnicity, parous, coffee, smoking, talc, BRCA1/2 mutation, medication use, gynaecologic surgery, cancer history Phase of menstrual cycle, age b35

Hospitalized

Original data

Mann–Whitney U test

Age, gender, menopausal status, renal or liver disease

Abbott Architect

Urban et al. [2]

Controls

66 ± 5.2 (mean ± SD)

101 (65 F, 34 Pre-M) 706

Trial

NA

Multiple regression models adjusting for all factors

Fujirebio ELISA

Bolstad et al. [18]

Healthy individualsc

48 (18–86)

1591 (801 F, fertility status NA)

Biobank sera

Log conversion

Univariate analysis/ multiple regression

Park et al. [13]

Healthy women Pregnant women

2182 (43% Pre-M) 72

Hospital employees

Original data

Mann–Whitney U test

Moore et al. [15]

Healthy women Pregnant women

1101 (41% Pre-M) 67

Trial

Log conversion

Fujirebio ELISA

CKD women Healthy women

113 (15% Pre-M) 68 (46% Pre-M)

Out-patients

Original data

Pearson's chi-square median test and Wilcoxon rank sum test Mann–Whitney U test

Age, menopausal status, pregnancy

Nagy et al. [20]

CKD

Abbott Architect

Hallamaa et al. [16]

54 126 778

Hospital patients

Original data

Tukey's test

Trial

Log conversion

Multiple regression

1515 (58% Pre-M)

Urban population

Original data

Braga et al. [19]

Healthy women

28 (50% Pre-M)

Hospital staff

Original data

Pearson's correlation analysis; ANOVA Multiple regression

Age, menstrual cycle, hormonal treatments, endometriosis Age, race, parous, TL, smoking, oral contraceptive use, history of breast cancer Age, menopausal status

Fujirebio ELISA

Yang et al. [21]

Healthy Pre-M with TL Endometriosis Healthy women at risk of ovarian cancer Healthy women

32 (27–39)d 31 (30–34)d NA (15–94) NA 72 (49–83) 44 (37–56) 34 (19–48) 52.3 ± 10.9 (mean ± SD) 46.1 ± 14.8 (mean ± SD) NA (25–68)

Age, BMI, race, family history of breast/ ovarian cancer, oral contraceptive use, nulliparous, history of endometriosis, current smoker, prior hysterectomy, current hormone therapy with intact uterus and prior hysterectomy Diurnal/seasonal variation, fasting/ non fasting, physical activity, gender, age, smoking, alcohol consumption, BMI, serum creatinine Age, pregnancy

Age, menopausal status

Roche Modular

Fujirebio ELISA and Abbott Architect

Abbott Architect

S. Ferraro et al. / Clinica Chimica Acta 438 (2014) 171–177

Urban et al. [12]

Fujirebio ELISA

Abbott Architect Roche Cobas

Post-M, post-menopausal women; BMI, body mass index; Pre-M, pre-menopausal women; NA, not available; CKD, chronic kidney disease; TL, laparoscopic tubal ligation. a Median (range), if not otherwise specified. b Factors associated with significant (P b 0.05) influence in each study are reported in italics. c Randomized for the assays. d Median (first to third quartiles).

173

174

S. Ferraro et al. / Clinica Chimica Acta 438 (2015) 171–177

population by decade of age, a statistically significant increase of HE4 concentrations was evidenced [15]. In general, pre-menopausal women appeared to exhibit HE4 concentrations significantly lower than post-menopausal women [21], but this difference became not significant when restricting the comparison between pre-menopausal and post-menopausal women b60 years [15,19]. On the other hand, Escudero et al. [17] were unable to show any significant increase in post- vs. pre-menopausal women. Lowe et al. by selecting a case series of post-menopausal women at high risk for OC confirmed the marked effect of age (P = 0.0001) [1]. Furthermore, in this cohort the effect of age at menarche was a significant (P = 0.03) predictor of HE4 concentrations [1]. Considering potential HE4 changes in response to different phases of menstrual cycle, Anastasi et al. [14] reported, on serial samples from 40 young healthy women (aged 30.8 ± 1.2 years) drawn at each phase. HE4 concentrations were lower in the follicular vs. the ovulatory phase (P b 0.0002). By partitioning subjects according to an age threshold of 35 years, the same evidence was however confirmed only in those b35 [14]. On the contrary, Hallamaa et al. [16] on a mixed population of 180 women (70% with endometriosis) did not find any statistically significant difference between HE4 concentrations in different phases of menstrual cycle. Similar results were obtained when subjects were further divided according to the use of hormonal medications [16]. HE4 concentrations were also shown to be significantly lower in pregnant vs. pre-menopausal women, without any influence by the pregnancy trimester [15]. Conversely, another study reported a slight but statistically significant increase of HE4 values in pregnancy with respect to healthy controls [13]. 3.1.3. Other factors Nagy et al. reported a statistically significant (P b 0.0001) increase of HE4 concentrations at the various stages of chronic kidney disease proportional to a decrease in the glomerular filtration rate (GFR) [20]. Renal failure has been reported as the most common cause of increased HE4 results in the clinical setting in the absence of malignant disease [17]. But already at physiologic GFR values, HE4 concentrations may be influenced by kidney function. Bolstad et al. [18] have reported that HE4 concentrations increase proportionally with serum creatinine concentrations. By assuming serum creatinine at 50 μmol/L as baseline, creatinine at 70 μmol/L was associated with 12% higher HE4 levels, 80 μmol/L with 17%, 90 μmol/L with 22% and 100 μmol/L with 27% higher levels [18]. Liver disease, in the absence of renal impairment, was also reported to increase HE4 concentrations [17]. Conversely, no differences in HE4 concentrations were detected between patients with endometriosis and healthy women [16]. HE4 concentrations were increased by 20–30% in smokers when compared with non smokers [2,12,18] and decreased with an increase in BMI [18]. HE4 concentrations were not influenced by circadian rhythm or seasonal variations and were not sensitive to alcohol consumption, fasting/non fasting status and physical activity before blood

sampling or hormone use [18]. No interference was reported for common chemotherapy drugs. Table 2 summarizes the most important biological and lifestyle factors influencing HE4 concentrations.

3.2. Factors influencing CA-125 concentrations in serum The main features of studies evaluating factors influencing CA-125 concentrations in serum are reported in Table 3. Females have higher CA-125 concentrations than males [17,22]. Discrepant data are available on the effect of age. Some authors did not report any correlation between age and CA-125 concentrations [16,22], whereas others found a statistically significant decrease of CA-125 concentrations in subjects ≥ 50 years [13,21]. The latter result was confirmed by Pauler et al. [9], who also showed a statistically significant effect of age at menarche and of age at the onset of menopause on marker concentrations. Conversely, Dehaghani et al. reported a higher rate of abnormal serum CA-125 results in the elderly (N 70 years) [23]. Some authors found higher CA-125 concentrations in premenopausal status [17,21]. Pauler et al. [9], by evaluating only postmenopausal women, evidenced significantly lower concentrations in those undergone hysterectomy. Controversial data are also available on the effect of the phases of menstrual cycle. Zweers et al. studied CA-125 in women stimulated for in vitro fertilization (IVF) and controls, during the first half of the cycle, at midcycle or at the moment of oocyte retrieval and at the second half of the cycle [24]. No significant changes of CA-125 concentrations were determined in normally cycling women by comparing luteal, follicular and peri-ovulatory phases, both supplemented or not with oral contraceptives. For women undergoing IVF, CA-125 concentrations were significantly higher in the luteal phase with respect to other phases. This evidence was not found in a subset of patients developing endometriosis [24]. On the contrary, Bon et al. showed statistically significantly higher serum CA-125 concentrations in the luteal phase and during menstruation vs. follicular and periovulatory phases [26]. Accordingly, under healthy conditions Kafali et al. reported mean CA-125 concentrations 22% higher during menstruation than those measured on other days (P b 0.001) [27]. For patients with endometriosis the CA-125 pattern was similar to healthy controls, but in the former group during menstruation CA-125 levels increased by ~200% with respect to other days [26]. Finally, Hallama et al. reported higher CA-125 values in secretory and proliferative phases only in women with endometriosis [16]. Other authors did not observe any cycle-dependent changes for CA-125 in healthy women [14,16,25]. By considering the possible relationship between CA-125 and luteinizing hormone (LH), follicle-stimulating hormone (FSH), prolactin, estradiol and progesterone concentrations, only Bon et al. observed a significant negative correlation between serum CA-125 and estradiol concentrations [26]. Pregnancy was reported to increase CA-125 concentrations [13].

Table 2 Synopsis of biological and lifestyle factors influencing HE4 concentrations in serum. Lowe et al. Anastasi et al. Escudero et al. Urban et al. Bolstad et al. Park et al. Moore et al. Nagy et al. Urban et al. Yang et al. [1] [14] [17] [2] [18] [13] [15] [20] [12] [21] Gender (female) Age Age at menarche Menstrual cycle (follicular phase) GFR Parous BMI Smoking Caffeine consumption

+a +

+ +

+

+ +

+

+

+

− +

+

+

+b + +

− +

b

GFR, glomerular filtration rate; BMI, body mass index. a +, positive association; −, negative association. b Statistically significant only in univariate linear regression (P = 0.03 and P = 0.04, respectively).

+

+

Table 3 Main features of studies related to biological factors and other characteristics influencing carbohydrate antigen 125 (CA-125) concentrations in serum. Authors

Population

Age (yr)a

Sample size

Setting of subjects

Data entering

Statistical model

Investigated factorsb

Zweers et al. [24]

Healthy women

NA

35

Volunteers

Original data

Phase of menstrual cycle, use of oral contraceptives

Bon et al. [26]

Healthy women

34.5 (28–38)

20

Original data

Erbagci et al. [25] Pauler et al. [9]

Healthy women Healthy Post-M

26.6 ± 0.9 (mean ± SD) 58 (41–60)

23 18,748

Infertility clinic out-patients Hospital staff Trial

Analysis of variance, Mann–Whitney U test, correlations Correlations

Original data Log conversion

Mann–Whitney U test Multiple regression

Kafali et al. [27]

26 (18–32) 27.5 (22–34) 60 (53–85)

12 16 203

Hospital patients

Log conversion

t-Test

Dehaghani et al. [23]

Healthy women Endometriosis Healthy Post-M

Phase of menstrual cycle, BMI Ethnicity, age, age of menarche, smoking, caffeine, previous history of hysterectomy, history of cancer, parity Phase of menstrual cycle, endometriosis

Out-patients

Original data

Chi-square

Bjerner et al. [22]

Healthy individuals

NA (18–86)

Biobank sera

Log conversion

Multiple regression

Lowe et al. [1]

Healthy Post-M at risk of ovarian cancer

55 (32–83)

250 (F) 250 (M) 155

Trial

Log conversion

Univariate analysis/ multiple regression

Anastasi et al. [14] Escudero et al. [17]

Healthy Pre-M Healthy individuals

31 (20–49) 52 (20–91)

University employees Hospitalized

Original data Original data

t-Test Mann–Whitney U test

Park et al. [13]

Healthy women Pregnant women

32 (27–39)c 31 (30–34)c

Hospital employees

Original data

Mann–Whitney U test

Age, pregnancy

Hallamaa et al. [16]

34 (19–48)

Hospital patients

Tukey's test

Age, menstrual cycle, hormonal treatments, endometriosis

52.3 ± 10.9 (mean ± SD)

Trial

Log conversion Log conversion

Multiple regression

Yang et al. [21]

Healthy Pre-M with TL Endometriosis Healthy women at risk of ovarian cancer Healthy women

40 101 (65 F, 34 Pre-M) 2182 (43% Pre-M) 72 54 126 778

Age, age of menarche, age at menopause, BMI, history of hormonal treatment, previous oral contraceptives, smoking Gender, age, physical activity, time period since last male, blood donor status, smoking habit, serum creatinine, BMI Age, age at menarche, height, weight, BMI, ethnicity, parous, coffee, smoking, talc, BRCA1/2 mutation, medication use, gynaecologic surgery, cancer history Phase of menstrual cycle, age b35 Gender, age, menopausal status, renal or liver disease

46.1 ± 14.8 (mean ± SD)

1515 (58% Pre-M)

Urban population

Original data

Braga et al. [19]

Healthy women

NA (25–68)

28 (50% Pre-M)

Hospital staff

Original data

Pearson's correlation analysis; ANOVA Multiple regression

Ethnicity, age, parous, TL, smoking, use of oral contraceptives, history of breast cancer Age, menopausal status

S. Ferraro et al. / Clinica Chimica Acta 438 (2014) 171–177

Urban et al. [12]

Phase of menstrual cycle

Age, menopausal status

NA, not available; BMI, body mass index; Post-M, post-menopausal women; F, females; M, males; Pre-M, pre-menopausal women; TL, laparoscopic tubal ligation. a Median (range), if not otherwise specified. b Factors associated with significant (P b 0.05) influence in each study are reported in italics. c Median (first to third quartiles).

175

176

S. Ferraro et al. / Clinica Chimica Acta 438 (2015) 171–177

Bjerner et al. reported that increasing serum creatinine concentrations, moderate alcohol consumption, active smoking and physical activity before blood sampling were not associated with increased CA-125 concentrations [22]. Conversely, Pauler et al. reported that smoking, caffeine consumption, race, as well as previous history of hysterectomy and cancer were significant predictors of CA-125 concentrations [9]. Dehaghani et al. confirmed the influence of smoking habit by reporting a CA-125 decrease in smokers in comparison to nonsmokers [23]. Finally, some authors showed that talcum powder use (with a positive sign), parous and hormone replacement therapy (with a negative sign) were significant predictors for CA-125 concentrations [1,16,23]. 3.3. Wiping out interfering factors to estimate HE4 biological variability The recognition of individual characteristics potentially affecting HE4 concentrations gains far more relevance when evaluating the biological variability (BV) of the marker in order to derive analytical goals and adequate interpretative criteria [19]. The definition of BV refers to the variation only due to “random, independent fluctuations of marker concentrations around the homeostatic set point” [28]. This definition implies to study BV by dealing with “ostensibly” healthy subjects, considered as being in a “steady-state condition” and not having any potential interfering factors. Noteworthy, the combination of subject features (e.g., age, smoking habit) at each individual level (e.g., age: N or b50 years; smoking: yes/no) characterizes the complex biology of each subject. The sum of these factors and their possible interactions can unpredictably modulate the pattern of biomarker release. In other words, the presence of one influencing factor (e.g., smoking) changes HE4 release with respect to absence of the interferent (no smoking) assumed as the reference condition. Here we have exemplified the concept by considering a dichotomous variable (two levels: yes/no), but most factors are multilevel (e.g., menstrual cycle phases, decade of age, etc.) and thus the interactions between all factors at different levels may characterize various and complex patterns of HE4 release. For instance, Bolstad et al. have shown that natural HE4 release pattern changes in each decade of age by assuming a different shape [18]. As a consequence, to evaluate BV components rigorous protocols have to be designed with strict criteria for subject selection in order to wipe out the effects of potential interfering conditions. Accordingly, in a recent study estimating the BV coefficients of HE4, we excluded subjects with history of any chronic disease and/or benign gynaecological or non-gynaecological disease, medication and oral contraceptive use, smoking habits or consumption of substantial quantities of alcohol [19]. Considering the pre-menopause subset, only women with a regular menstrual cycle were included. To further exclude any potential influence of menstrual phases, blood drawing was performed between the 12th and 14th days of the menstrual cycle, corresponding to the ovulation period. In addition, the time interval between serial samples drawing (monthly) and the time length of the study (four consecutive months) were standardized for all individuals. By comparing the results by Braga et al. [19] with those of previous studies performed only on CA125 [29–31], we could speculate on the bias introduced in the evaluation of CA-125 BV when interfering factors were partially or not excluded at all. In those studies the inflation of BV coefficients is likely associated with sample collection partially performed during menstruation, inclusion of individuals with various pathological conditions (e.g., carrying malignancies or suffering renal impairment) or to the blending of results from subjects in pre- and post-menopausal status. In relation to these factors, CA-125 concentrations may increase and cover wide range of results with unpredictably expanding effects on biological CV estimates. In the absence of potential interfering factors, Braga et al. reported a strong difference between pre- and post-menopausal status only for HE4 intra-individual BV and not for marker concentrations [19]. This result appears in line with previous data showing that menopause per se

does not affect HE4 concentrations and thus should not be definitively considered as influencing status. Conversely, there is evidence that the biological behaviour of HE4 relevantly changes from pre- to postmenopausal period by assuming a specific pattern of release in each condition. This implies the need to differently manage pre- and postmenopausal women when interpreting changes in marker concentrations by the adoption of different reference change values (RCV) according to the hormonal status of the individual [19]. 4. Discussion In the OC framework the poor specificity of available biomarkers has historically conditioned the field of application and penalized their cost–effectiveness ratio. Now that the more specific HE4 is available for use in clinical practice, there is still confusion on the way to manage marker results [4]. Authors have often assumed that factors influencing CA-125 release are likely to influence HE4 concentrations in serum. Accordingly, the use of RI partitioned for menopausal status and of diagnostic algorithms including both CA-125 and HE4 values adjusted for menopausal status has been recommended [8,32]. On the other hand, the practical impact of the recommendations released by the U.K. National Institute for Health and Clinical Excellence (NICE) on biomarker use for OC detection in primary and secondary care has recently fostered the relevance of considering interfering conditions on marker increase [33]. In face of the tidal wave of false positive CA-125 results, the appraisal of interfering biological and lifestyle factors has appeared as an aid in improving the interpretation of marker increases and/or in suggesting the application of specific diagnostic algorithm in selected categories of patients. For instance, in implementing the NICE algorithm some institutions have proposed to consider CA-125 increases over the diagnostic threshold of 35 kU/L as clinically relevant only when detected in patients N 50 years [34]. The adjustment of biomarker concentrations for potential influencing factors can be performed only whether robust evidence of their effect on marker release is available. By reviewing studies on biological and lifestyle characteristics influencing serum HE4 (and CA-125) concentrations, we however retrieved uncorroborated and often contrasting data. This is likely due to the wide heterogeneity of the study protocols, often including small numbers of subjects with respect to the number of studied variables, sometimes further partitioned in various subsets (e.g., four menstrual phases). The use of pair-wise statistical comparisons between different groups instead of more robust regression analyses, by adjusting each factor for other variables, may further affect the statistical power of analyses providing biased information. By considering the menopausal status, only two studies reported a significant increase of serum CA-125 concentrations in pre- vs. postmenopausal women, although this evidence was quite modest owing to the heterogeneous population studied and/or to the simplistic statistical approach [17,21]. Thus, if CA-125 elevations have different clinical relevance in patients younger vs. older than 50 years [9,13], this assumption should not necessarily be extended to manage subjects in pre- or post-menopausal status differently. Theoretically, there is no background to adjust CA-125 for menopausal status, although we could reasonably speculate on the wider heterogeneity of CA-125 values in pre- vs. post-menopausal women associated with a higher prevalence of overt/subclinical benign gynaecological disease in the first case. Some factors reported to strongly influence CA-125 concentrations produce no or minimal effect on HE4. In particular, we refer to phases of menstrual cycle and endometriosis. There is only one small study showing that the ovulation phase appears to increase HE4 concentrations in women younger than 35 years [14]. In comparison with CA-125, HE4 does not exhibit elevations in endometriosis. Indeed, HE4 has been proposed as the marker of choice for distinguishing endometriosis from OC [35]. Although different decisional levels and algorithms have been proposed to interpret HE4 results in pre- and post-menopausal women in

S. Ferraro et al. / Clinica Chimica Acta 438 (2014) 171–177

the clinical setting, we were unable to find definitive evidence supporting changes of marker concentrations in relation to menopausal status. On the contrary, the most robust evidence, meeting the agreement of all retrieved studies, concerned the effect of age increase on HE4 release. Surprisingly, this relationship appears to be non-linear and quite complex, described by a higher degree polynomial [18]. In particular, the HE4 release changes with an increasing trend for each decade of age. If we consider the 20–30 years decade as baseline and women in the range of 50–64 years of age at higher risk for OC, the percent increase of HE4 by decade should be of help in interpreting marker values. With respect to 20–30 years decade, healthy women of 60 years of age would show an HE4 increase of 37% [18]. This evidence, quite robust for the study design, the sample size and the employed statistical approach, suggests that use of RI when partitioned for menopausal status is not reliable for HE4 interpretation. The concept of the limited utility attributable to RI has been recently reinforced by Braga et al., who showed that, according to HE4 BV, the adoption of the RCV concept represents the most reliable interpretative criteria for HE4 [19]. Amongst other factors, the smoking significantly influences HE4 circulating concentrations and is associated with a strong HE4 expression in bronchial epithelium, already present under physiologic conditions [5]. By considering renal function, there is evidence on the effect of GFR on HE4 concentrations according to a linear relationship. This could become relevant when interpreting marker concentrations in the follow-up of OC patients, since decreases in GFR are often observed in aged patients and in those that have undergone chemotherapy [36]. In addition to a reduced renal clearance, circulating HE4 concentrations might increase in renal disease because of an increased HE4 expression by the injured kidney tissue itself [37]. Interestingly, in males, circulating HE4 concentrations are lower than in women, but the age-related increase is more pronounced. Since HE4 seems to exert an inhibitory role on prostate-specific antigen (PSA), possibly the age-related increase of PSA might be physiologically correlated to that of HE4 [38]. Finally, the lack of data on assay comparability does not permit marker adjustment from studies employing different assays, as the data pooling might be censored, with a consequent loss of the accumulated clinical experience [4]. In conclusion, the literature on the biological and lifestyle factors potentially influencing serum HE4 concentrations is scarce and the available knowledge needs to be integrated with further studies. From available findings, factors, such as age N 55 years, GFR and smoking, seem to markedly influence HE4 concentrations. Although no variations in HE4 levels in serum can be clearly associated to menopausal status, a strong difference in biomarker intra-individual BV due to the fertility status has been reported. According to this difference, one should differently interpret HE4 individual changes in pre- and post-menopausal women. References [1] Lowe KA, Shah C, Wallace E, et al. Effect of personal characteristics on serum CA125, mesothelin, and HE4 levels in healthy post-menopausal women at high-risk for ovarian cancer. Cancer Epidemiol Biomarkers Prev 2008;17:2480–7. [2] Urban N, Thorpe JD, Bergan LA, et al. Potential role of HE4 in multimodal screening for epithelial ovarian cancer. J Natl Cancer Inst 2011;103:1630–4. [3] Nolen B, Velikokhatnaya L, Marrangoni A, et al. Serum biomarker panels for the discrimination of benign from malignant cases in patients with an adnexal mass. Gynecol Oncol 2010;117:440–5. [4] Ferraro S, Panteghini M. Is serum human epididymis protein 4 ready for prime time? Ann Clin Biochem 2014;51:128–36. [5] Jiang SW, Chen H, Dowdy S, et al. HE4 transcription and splice variant specific expression in endometrial cancer and correlation with patient survival. Int J Mol Sci 2013;14:22655–77. [6] Galgano MT, Hampton GM, Frierson Jr HF. Comprehensive analysis of HE4 expression in normal and malignant human tissue. Mod Pathol 2006;19:847–53. [7] Bandiera E, Romani C, Specchia C, et al. Serum human epididymis protein 4 and risk for ovarian malignancy algorithm as new diagnostic and prognostic tools for epithelial ovarian cancer management. Cancer Epidemiol Biomarkers Prev 2011;20: 2496–506.

177

[8] Moore RG, McMeekin DS, Brown AB, et al. A novel multiple bioassay utilizing HE4 and CA125 for the prediction of ovarian cancer in patients with a pelvic mass. Gynecol Oncol 2009;112:40–6. [9] Pauler DK, Menon U, McIntosh M, Symecko HL, Skates SJ, Jacobs IJ. Factor influencing serum CA125II levels in healthy postmenopapausal women. Cancer Epidemiol Biomarkers Prev 2001;10:489–93. [10] McIntosh MW, Urban N, Karlan B. Generating longitudinal screening algorithms using novel biomarkers for disease. Cancer Epidemiol Biomarkers Prev 2002;11: 159–66. [11] Ferraro S, Braga F, Lanzoni M, Boracchi P, Biganzoli EM, Panteghini M. Serum human epididymis protein 4 vs carbohydrate antigen 125 for ovarian cancer diagnosis: a systematic review. J Clin Pathol 2013;66:273–81. [12] Urban N, Thorpe J, Karlan BY, et al. Interpretation of single and serial measures of HE4 and CA-125 in asymptomatic women at high risk for ovarian cancer. Cancer Epidemiol Biomarkers Prev 2012;21:2087–94. [13] Park Y, Kim Y, Lee EY, Lee JH, Kim HS. Reference ranges for HE4 and CA125 in a large Asian population by automated assays and diagnostics performances for ovarian cancer. Int J Cancer 2012;130:1136–44. [14] Anastasi E, Granato T, Marchei GG, et al. Ovarian tumor marker HE4 is differently expressed during the phase of menstrual cycle in healthy young women. Tumour Biol 2010;31:411–5. [15] Moore RG, Miller MC, Eklund EE, et al. Serum levels of the ovarian cancer biomarker HE4 are decreased in pregnancy and increase with age. Am J Obstet Gynecol 2012; 206:349.e1–7. [16] Hallamaa M, Suvitie P, Huhtinen K, et al. Serum HE4 concentration is not dependent on menstrual cycle or hormonal treatment among endometriosis patients and healthy premenopausal women. Gynecol Oncol 2012;125:667–72. [17] Escudero JM, Auge JM, Filella X, et al. Comparison of serum human epididymis protein 4 with cancer antigen 125 as a tumor marker in patients with malignant and nonmalignant diseases. Clin Chem 2011;57:1534–44. [18] Bolstad N, Oijordsbakken M, Nustad K, Bjerner J. Human epididymis protein 4 reference limits and natural variation in a Nordic reference population. Tumour Biol 2012;33:141–8. [19] Braga F, Ferraro S, Mozzi R, Panteghini M. The importance of individual biology in the clinical use of serum biomarkers for ovarian cancer. Clin Chem Lab Med 2014. http://dx.doi.org/10.1515/cclm-2014-0097 [Ahead of print]. [20] Nagy Jr B, Krasznai ZT, Balla H, et al. Elevated human epididymis protein 4 concentrations in chronic kidney disease. Ann Clin Biochem 2012;49:377–80. [21] Yang J, Sa M, Huang M, et al. The reference intervals for HE4, CA125 and ROMA in healthy female with electrochemiluminescence immunoassay. Clin Biochem 2013; 46:1705–8. [22] Bjerner J, Hogetveit A, Wold AK, et al. Reference interval for carcinoembryonic antigen (CEA), CA125, MUC1, alfa-foeto-protein (AFP), neuron-specific enolase (NSE) and CA19.9 form the NORIP study. Scand J Clin Lab Invest 2008;68:703–13. [23] Dehaghani AS, Ghiam AF, Hosseini M, et al. Factor influencing serum concentrations of CA125 and CA15-3 in Iranian healthy post-menopausal women. Pathol Oncol Res 2007;13:360–4. [24] Zweers A, De Boever J, Serreyn R, et al. Correlation between peripheral CA-125 levels and ovarian activity. Fertil Steril 1990;54:409–14. [25] Erbagci AB, Yilmaz N, Kutlar I. Menstrual cycle dependent variability for serum tumor marker CEA, AFP, CA19.9, CA-125 and CA15-3 in healthy women. Dis Markers 1999;15:259–67. [26] Bon GG, Kenemans P, Dekker JJ, et al. Fluctuation in CA-125 and CA15-3 serum concentrations during spontaneous ovulatory cycles. Hum Reprod 1999;14:566–70. [27] Kafali H, Artuc H, Demir N. Use of CA-125 fluctuation during the menstrual cycle as a tool in the clinical diagnosis of endometriosis; a preliminary report. Eur J Obstet Gynecol Reprod Biol 2004;116:85–8. [28] Fraser CG, Harris EK. Generation and application of data on biological variation in clinical chemistry. Crit Rev Clin Lab Sci 1989;27:409–37. [29] Browning MCK, McFarlane NP, Horobin JM, et al. Objective interpretation of results for tumour marker. J Nucl Med Allied Sci 1990;34:89s–91s. [30] Tuxen MK, Soletormos G, Petersen PH, et al. Assessment of biological variation and analytical imprecision of CA-125, CEA, and TPA in relation to monitoring of ovarian cancer. Gynecol Oncol 1999;74:12–22. [31] Trape J, Perez de Olaguer J, Buxo J, et al. Biological variation of tumor markers and its application in the detection of disease progression in patients with non-small cell lung cancer. Clin Chem 2005;51:219–22. [32] Plebani M, on behalf of the HE4 Study Group. HE4 in gynecological cancers: report of a European investigators and experts meeting. Clin Chem Lab Med 2012;50: 2127–36. [33] National Collaborating Centre for Cancer (UK). National Institute for Health and Clinical Excellence (NICE): Guidance. Ovarian cancer: the recognition and initial management of ovarian cancer. Cardiff, UK: National Collaborating Centre for Cancer; 2011. [34] Finlayson C, Tucker C. Lothian Primary Care CA125 Guidance Notes. Available at http:// www.scan.scot.nhs.uk/Documents/CA125%20Guidance%20Notes.pdf; February 2014. [Accessed June 2014]. [35] Kadija S, Stefanovic A, Jeremic K, et al. The utility of human epididymal protein 4, cancer antigen 125, and risk for malignancy algorithm in ovarian cancer and endometriosis. Int J Gynecol Cancer 2012;22:238–44. [36] Gizzo S, Ancona E, Saccardi C, D'Antona D, Nardelli GB, Plebani M. Could kidney glomerular filtration impairment represent the “Achilles heel” of HE4 serum marker? A possible further implication. Clin Chem Lab Med 2014;52:e45–6. [37] LeBleu VS, Teng Y, O'Connel JY, et al. Identification of human epididymis protein 4 as a fibroblast-derived mediator of fibrosis. Nat Med 2013;19:227–31. [38] Schiumarini D, Ferraro S, Panteghini M. Biochemical and functional characteristics of human epididymis protein 4 (HE4). Biochim Clin 2014;38:368–73.