Development & Validation of a Sensitive Assay for the Detection of 5α-Dihydrotestosterone in Serum with LC-ESI-MS/MS

(1)

MSc Chemistry

Track Analytical Sciences

Master Thesis

Development & Validation of a Sensitive Assay for the

Detection of 5α-Dihydrotestosterone in Serum with

LC-ESI-MS/MS

by

Ditte Bijlmakers

UvA: 11812060

VU: 2630062

October

2019

54 ECTs

February – October 2019

Supervisor/Examiner: Examiner:

(2)

1. Abstract

Background: In the prostate, testosterone (Tsto) is reduced into 5α-dihydrostestosterone (DHT). The

androgen receptor (AR) is the main driver in prostate cancer and is activated by androgen hormones. In prostate cancer, AR signaling is often deregulated. This deregulation promotes prostatic cells to grow, resulting in expression of oncogenes. Prostatic cells promote androgen activity and express higher DHT levels binding to the AR. AR signaling can be blocked by hormonal therapy using abiraterone or enzalutamide, resulting in decreased androgen levels. Therefore, in serum from hormonal treated patients, circulating DHT levels are significantly low and detected in relatively low concentrations (nmol/L or pmol/L). The relationship between serum DHT levels and prostate cancer is poorly understood, and therefore an interesting target for clinical research.

Method: We report a LC-MS/MS (UPHLC with QTRAP6500+ from Sciex) based assay with

electrospray ionization in positive-mode. For this assay development, left over serum samples were used. This assay used 250 µL serum and a deuterated internal standard, and involved liquid-liquid extraction (LLE) with hexane/ethyl acetate (4:1 v/v) during sample preparation. In addition, derivatization with either 2M hydroxylamine or 2%-O-tert-butyl hydroxylamine (OTB) g/mL, or no derivatization step was used. Separation was performed using either a C18 (1.7 or 2.6 µm particles) or phenyl hexyl column in reversed phase. Validation criteria were based on CLSI guidelines.

Results: Three LC-methods were developed for the detection of DHT without derivatization or either

hydroxylamine or OTB derivatization. With these three assays a method comparison was performed. The LC-method using OTB derivatization was able to detect DHT in serum from healthy men, pre- and post-menopausal women, and hormonal treated men. In contrast to other DHT-assays using different (or no) derivatization reported (lower) limits of quantification (LLOQ’s) between 0.0215 and 0.250 nmol/L, we report a LLOQ of 0.011 nmol/L with the OTB assay. Possible interferences were negligible in serum from healthy- men and women, and hormonal treated men. This assay showed linearity with R2_{=0.9924 and a method imprecision of 7%.}

Conclusion: This assay can be adopted for measurements of DHT in healthy- and diseased men and

women. To assess the relationship between DHT concentrations of males treated with hormonal therapy and clinical values related to prostate cancer, further clinical research has to be done. This research includes the application of this assay to investigate whether DHT is useful for monitoring hormonal therapy in prostate cancer patients.

(3)

2. Abbreviations

AA = anti-androgen

ADT = androgen deprivation therapy

APCI = atmospheric pressure chemical ionization APPI = atmospheric pressure ionization

AR = androgen receptor CE = collision energy

CRPC = castration-resistant prostate cancer CV = method imprecision DHEA = dehydroepiandrosterone DHT = 5α-dihydrotestosterone DHT-d3 = 5α-dihydrotestosterone-16,16,17-d3 DMSO = dimethylsulfoxide DP = declustering potential EpiT = epitestosterone ESI = electrospray ionization FBS = fetal bovine serum

FSH = follicle-stimulating hormone GnRH = gonadotropin-releasing hormone HA = hydroxylamine

LC = liquid chromatography

LC-MS/MS = liquid chromatography tandem mass spectrometry

LC-ESI-MS/MS = liquid chromatography electrospray ionization tandem mass spectrometry LH = luteinizing hormone

LHRH = luteinizing hormone-releasing hormone LLE = liquid-liquid extraction

(L)LOQ = (lower) limit of quantification

m/z = mass-to-charge ratio

MRM = multiple reaction monitoring MS = mass spectrometer

MS/MS = tandem mass spectrometry N/D = not detected

OTB = O-tert-butyl hydroxylamine PSA = prostate specific antigen

psi = pound per square inch (pressure) QC = quality control RIA = radioimmunoassay S/N = signal-to-noise SD = standard deviation tR = retention time Tsto = testosterone

UHPLC = ultra-high performance liquid chromatography V = volt

(6)

3. Introduction

3.1 Prostate Cancer and The Role of 5α-Dihydrotestosterone

Worldwide, prostate cancer is the most frequently diagnosed malignant cancer in male patients. In 2016, 1.4 million men were diagnosed with prostate cancer. In 2018, prostate cancer was diagnosed in over one-half of the countries in the world. Prostate cancer increases rapidly with age and causes almost 3% of deaths in men older than 55 years [1,2].

Steroid hormones play a major role in the progression of prostate cancer, and are controlled by the hypothalamus and the pituitary gland [3,4]. When hormone levels are decreased, the hypo-thalamus responses by releasing the gonadotropin-releasing hormone (GnRH). GnRH travels via the blood system to the hypothalamus, and binds to plasma membrane receptors on pituitary gonadotrophic cells. In addition, the hypothalamus releases the luteinizing hormone (LH) and the follicle-stimulating hormone (FSH). LH and FSH regulate the function of the testis and the ovaries, including the semen production in men, and the regulation of the menstrual cycle in women. When LH and FSH enter the general blood circulation, the production of androgens in the testes, such as testosterone (Tsto) and 5α-dihydrotestosterone (DHT), is stimulated (Figure 1) [4–8].

The main driver of prostate cancer is the androgen receptor (AR). The AR is a nuclear receptor that regulates the growth of the prostate and functions as a transcription factor [9,10]. The AR is activated by androgen hormones. The production of androgens is derived from a process primarily starting with cholesterol (Figure 2) [10]. Androgens can promote cell division and stimulate the development of carcinogenesis [11]. Therefore, androgens play a significant role in the development of prostate cancer. Tsto is endogenously synthesized by Leydig cells. These cells start the production of Tsto in the testicles in men and ovaries in women [3]. In the prostate, Tsto is converted into DHT via the 5α-reductase enzyme. This reaction includes a reduction involving Nicotinamide Adenine Dinucleotide Phosphate (NADPH) that is reduced into Nicotinamide Adenine Dinucleotide (NADP+₎

(Figure 3). In healthy cells, DHT is responsible for the development of the genitals during puberty and prostate growth. However, prostatic cells promote androgen activity and express higher DHT levels that bind to the AR, resulting in deregulation of AR signaling. DHT has more binding affinity to the AR than Tsto, considering DHT as the most relevant androgen in prostate cancer. Due to this deregulation, AR dependent signaling can form an auto-regulatory negative feedback loop, where noncoding RNA stimulates the prostate cell to grow due to transactivation through direct gene binding [12]. During this cell growth, the AR is translocated to the nucleus of the cell and drives the expression of oncogenes which results in proliferation of cancer cells in the prostate [3,12]. This results in a stimulation of gene expression, generating over-signaling, and ultimately growth of cancer cells.

Figure 1 – Production of androgen (DHT) by interactions with the hypothalamus [5]

(7)

Figure 2 –Biosynthesis pathway of steroid hormones [13]

Figure 3 – Reduction mechanism of testosterone with 5α-reductase [14]

3.2 Treatments in Prostate Cancer

First line treatment in advanced prostate cancer is androgen deprivation therapy (ADT) [6]. ADT decreases the formation of DHT from testicular androgens [7]. There are two main types of hormone therapy in this treatment. The first therapy uses an agonist (GnRH) that lowers circulating Tsto levels through medical castration, also known as luteinizing hormone-releasing hormone (LHRH) [3,4]. The second type of therapy is with an anti-androgen (AA), which competes with Tsto at the AR in the prostate cell, and stops androgens from working [6]. Abiraterone is commonly used to decrease adrenal androgens and indirectly inhibits the AR signaling pathway that stops the production of Tsto [8]. Another commonly used AR blocker is enzalutamide. This AR blocker binds to the ligand-binding site of the AR, and stops the AR from working. This binding process prevents nuclear translocation of the AR and DNA binding [8].

(8)

3.3 DHT in (Chemically) Castrated Men

After (chemical) castration, the AR responds with alterations (molecular and biochemical), causing hypersensitivity to low ligand levels in the prostate and in cancer cells by in vitro and in vivo changes [15]. Another mechanism causing hypersensitivity, is the response of the AR to low androgen levels due to decreased Tsto levels. Low androgen levels are a common characteristic of resistance to chemical castration, termed as castration-resistant prostate cancer (CRPC). In this case, the AR is more sensitive to DHT binding, to compensate for low androgen levels [16]. Therefore, cancer cells grow faster than healthy cells, which ensures transactivation in CRPC. This means that CRPC may be androgen-dependent [15,16].

After chemical castration, DHT production in the prostate is reduced. Reference DHT values for healthy man have been published by Mayo Clinical Medical Laboratory (Rochester, MN), and lie between 0.38 and 3.27 nmol/L [17]. These values are lower after hormonal treatment with abiraterone or enzalutamide (<0.30 nmol/L) [3,18]. DHT levels in healthy women differ from that in man, since women have a lower Tsto production. These DHT values also vary between pre- and post-menopausal women. For pre-post-menopausal women, DHT concentrations lie around 0.3 nmol/L, and 0.1 nmol/L for post-menopausal women [19]. This illustrates that DHT is significantly detected in relatively low concentration, mostly between 1 and 5 nmol/L or even in picomolar ranges.

3.4 Current DHT Quantification

In the clinical field, measurement of Tsto and DHT is commonly performed using liquid chromatography tandem mass spectrometry (LC-MS/MS) [20]. LC-MS/MS is preferred over (radio)-immunoassays (RIA), since LC-MS/MS reaches lower limits of quantification (LLOQ), and has more selectivity [21]. In addition, immunoassays often require large sample volumes [3,22]. Mostly, electrospray ionization (ESI), atmospheric pressure photo ionization (APPI), or atmospheric pressure chemical ionization (APCI) are used in positive ion mode coupled to a triple quadrupole MS. For sample preparation, liquid-liquid extraction (LLE) or on-line solid phase extraction (SPE) is commonly used.

Due to low proton affinity, DHT has an extremely poor response in ionization methods as APPI, APCI and ESI. Furthermore, DHT has no conjugated structure, and produces no specific fragment ions of high abundance [23]. Therefore, derivatization reagents are commonly used to improve these limitations. There are methods that are able to detect DHT without using a derivatization reagent. In these assays, fragment ions due to water loss from the protonated molecular ions are used [23]. Unfortunately, methods without derivatization reagents mostly lack sensitivity and selectivity for measuring DHT at low concentrations in serum as observed in (chemical) castrated men.

Using a derivatization reagent has the disadvantage that it is more time consuming. On the other hand, derivatization increases sensitivity, and the ionization potential of steroid molecules [20]. Some commonly used derivatization reagents are hydroxylamine (HA), and 2-picolinic acid with 4-(dimethylamino) pyridine (DMAP) and 2-methyl-6-nitrobenzoic anhydrine (MNBAn) [23,24]. Another oxime derivatization reagent is O-tert-butyl hydroxylamine (OTB). OTB reacts similar to hydroxyl-amine, but provides better ionization due to its structure.

(9)

3.5 Applications and Research Question

Androgenic activity plays a major role in processes in the prostate. DHT is considered to be the most potent agonist binding to the AR. However, the relationship between serum DHT levels and prostate cancer is poorly understood. Therefore, DHT is an interesting target for clinical research. Consistently, steroid concentrations are relatively low in serum which requires development of a sensitive LC-MS/MS-based assay [25]. The aim of this study is to develop a sensitive and accurate assay for the detection of DHT in serum – Since various ADT treatments affect the DHT concentration in the prostate [26]. This can give insight into cellular mechanisms of hormonal treatments. In the context of various research questions and potentially for clinical questions, it is relevant to quantify DHT in serum of males, pre- and post-menopausal females giving insight in reference DHT values.

4. Material and Methods

4.1 Reagents

Dihydrotestosterone (DHT), dihydrotestosterone-16,16,17-d3 (DHT-d3), testosterone (Tsto),

epi-testosterone (EpiT), androstenedione, 17-Hydroxy-progesterone (17-HOP), dehydroepiandrosterone (DHEA), cortisol, estrone (E1), 17β-estradiol (E2), hydroxylamine hydrochloride, O-tert-butyl hydroxylamine hydrochloride (OTB), acid washed charcoal powder, and bilirubin (≥98%), were obtained from Sigma-Aldrich (St Louis, MO, USA). Dextran (clinical grade) was obtained from MP Biomedicals (Solon, OH, USA). Dimethyl sulfoxide (DMSO), hexane, ethyl acetate, methanol, and ethanol (≥98%) were from highest analytical grade. Physiological saline (0.9%-NaCl) was used for analysis of reference standards.

4.2 Samples for Assay Development

Obtained serum samples were stored at 4⁰C in a rapid separator tube (RST), and kept for a maximum of 7 days. Left-over serum samples used for this assay development were from randomly chosen (anonymous) patients from the Antoni van Leeuwenhoek hospital (The Netherlands, Amsterdam). 4.3 LC-MS/MS Conditions

For the detection of DHT, a one-dimensional Shimadzu Nexera LC-30AD UHPLC system (AB Sciex, Concord, ON, Canada) was used with either a reversed phase 2.6 µm or 1.7 µm C18 column (50 x 2.1 mm, 100 Å internal diameter, Phenomenex , Torrance, CA USA), or a reversed phase 1.7 µm phenyl hexyl column (50 x 2.1 mm, 100 Å internal diameter, Phenomenex , Torrance, CA USA). The LC system was connected to a quadrupole MS QTRAP 6500+ (AB Sciex, Concord, ON, Canada). The mobile phase was established by combining two solvents. Solvent A containing water with 0.1% formic acid (v/v) to improve protonation. Solvent B contained methanol, and was used with a gradient to elute the analyte. The flow of the mobile phases was regulated by pumps. The pumps provided accurate flow rates through the system and were connected with switching valves controlling the direction of the flow. Pump A was connected to a mixer and a ten port switching valve. The ten port valve was connected with the mixer of pump B and the autosampler. The autosampler was connected to a six port valve. When a sample was injected, the valve switched, and connected to the column where the sample was eluted into the ionization source (ESI+) of the mass spectrometer (MS).

(10)

4.4 Internal Standard Mixture

In this assay we used 5α-dihydrotestosterone-16,16,17-d3 (DHT-d3) from CDN isotopes

(Pointe-Claire, QC, Canada) as internal standard for DHT. DHT-d3 is a stable labelled nonradioactive isotope of

DHT containing three deuterated groups [27]. The internal standard solution was prepared by diluting a stock solution of 10 mM DHT-d3 (in DMSO) with methanol giving a concentration of 25

nmol/L.

4.5 Preparation of Reference Standards

We developed three LC-methods for the detection of DHT from serum, namely: underivatized, derivatization with either hydroxylamine or OTB. Calibrators for each method were prepared in methanol. Physiological saline (0.9%-NaCl) was used as matrix, because 0.9%-NaCl has similar physiological properties as serum, and matches the osmolarity of the human body. The measurement range differs for all three methods, therefore different calibrators were used (Table 1). Calibrators were diluted from a 10 µM DHT solution derived from a stock solution of 10 mM DHT (in DMSO). A weighting factor of 1/x2_{was used, as the absolute variation for high concentrations in this}

linear regression is relatively higher than for low concentrations. All three methods showed linearity (R2_{>0.99) in the calibrations curves.}

Table 1 – DHT calibrators in nmol/L for method underivatized, derivatization with hydroxylamine (HA), and derivatization with O-tert-butyl hydroxylamine (OTB)

Calibrator Underivatized [DHT] in nmol/L HA [DHT] in nmol/L OTB [DHT] in nmol/L C8 - - 10 C7 20 15 5 C6 15 10 2.5 C5 10 5 1 C4 5 2.5 0.5 C3 2.5 1 0.1 C2 1 0.5 0.025 C1 0.5 0.1 0.005

4.6 Protocol Charcoaling of Serum

To create a steroid depleted matrix similar to the human body, fetal bovine serum (FBS) was stripped by addition of dextran and activated charcoal powder (1:10 w/w). Dextran coated charcoal absorbs free steroids in serum, and dextran was added to prevent protein loss. Serum with dextran coated charcoal was gently mixed overnight on a shaker table at 4°C. Charcoal was removed from serum by centrifugation and removal of the supernatant. This procedure was performed twice to ensure removal of all steroids.

4.7 Sample Preparation

Serum samples were prepared by mixing 250 µL serum with 10 µL of the internal standard mixture in 3 mL glass test tubes (i.d. 10.5 mm). The organic extract was obtained by addition of 1 mL hexane/ ethyl acetate (4:1 v/v). The samples were shaken for 15 minutes at 480 rpm on an orbital shaker (IKA, Staufen, Germany), and centrifuged (Eppendorf, Hamburg, Germany) for 5 minutes at 5,000 rcf. Next, the aqueous phase was snap frozen in dry ice with alcohol, and the organic phase was collected in 1.5 mL glass vials (Phenomenex, Torrance, CA, USA). The organic phase was dried in a SpeedVac (Thermo Fisher Scientific, Waltham, MA, USA), and the solid was reconstituted in either 100 µL methanol/water (1:1 v/v), or 100 µL of 2.0 M hydroxylamine hydrochloride in methanol/water (1:1

(11)

v/v) and heated at 60°C for 30 minutes, or 100 µL of 2%-OTB (g/mL) in methanol/water (1:1 v/v). The samples were span down at high speed, and a volume of 50 µL was injected for LC-MS/MS analysis. 4.8 DHT Derivatization with Hydroxylamine and O-tert-butyl hydroxylamine

Hydroxylamine hydrochloride is commonly used to derivatize keto-steroids. Hydroxylamine is an inorganic compound reacting as a reducing agent. The DHT-oxime of hydroxylamine is illustrated in

Figure 4. The DHT-oxime has one abundant parent ion at m/z 306. Unfortunately, there is no specific

fragment ion for the DHT-oxime of hydroxylamine. Therefore, the DHT-oxime of hydroxylamine was tuned with the MS for the most abundant fragment ion. Figure 5 shows the DHT-oxime of OTB. OTB has a similar structure to hydroxylamine, containing three extra methyl-groups. It is known that the DHT-oxime of OTB has only one specific fragment ion, namely m/z 362 > 306. This fragment ion is derived from neutral loss of the O-tert-butyl moiety [23].

Figure 4 – DHT-oxime of hydroxylamine [28] Figure 5 – DHT-oxime of O-tert-butyl hydroxylamine [23]

4.9 Gradient Methods for the Detection of DHT

For the detection of DHT, four gradient methods were used (Table 2). Gradient 1 was used to detect underivatized DHT, and gradient 2 to detect the DHT-oxime of hydroxylamine. The detection of the DHT-oxime of OTB started with gradient 3, and after optimization we used gradient 4.

Table 2 – Gradient LC-methods for the detection of DHT

1 - Gradient for the detection of underivatized DHT

2 - Gradient for the detection of derivatized DHT with hydroxylamine

3 - Gradient for the detection of derivatized DHT with OTB

4 - Optimized gradient for the detection of derivatized DHT with OTB

(12)

4.10 Ion Transitions and MRM Settings

The analyte, internal standard, and its derivatives were tuned and the MS parameters were optimized using flow injection analysis, where the sample was directly injected into the MS. The optimized instrument settings were as follows: collision gas was set at medium, ion spray voltage was 5500 V, curtain gas was set at 40 psi, source temperature at 450⁰C, nebulizing gas at 60 psi and Turbo Ion Spray gas at 40 psi.

In tandem MS (MS/MS), mostly selected reaction monitoring, also termed as multiple reaction monitoring (MRM) is used. MRM means that within one time scale more than one fragment ion of a selected precursor ion can be measured. In MS/MS this is performed in Q1 and Q3. The optimal MRM settings and its transitions for underivatized DHT, and DHT-oximes are listed in Table 3.

Table 3 – MRM settings for underivatized DHT and derivatization with hydroxylamine (HA) and OTB (DP = declustering potential, EP = entrance potential, CE = collision energy, CXP = collision cell exit potential, Da = Dalton; msec = mili-seconds; V = volt)

Compound Name Q1 Mass (Da) Q3 Mass (Da) Dwell Time (msec) DP (V) EP (V) CE (V) CXP (V) DHT 291.226 255.100 60.0 191.000 10.000 21.000 14.000 DHT-d3 294.189 258.200 60.0 211.000 10.000 21.000 16.000 DHT-HA 306.256 105.200 60.0 151.000 10.000 51.000 12.000 DHT-d3-HA 309.291 77.000 60.0 146.000 10.000 103.000 8.000 DHT-OTB 362.030 306.300 60.0 111.000 5.000 25.000 5.000 DHT-d3-OTB 365.016 309.100 60.0 91.000 5.000 40.000 12.000 4.11 Statistical Analysis

During assay development, we performed a method comparison using a Passing-Bablok regression. This analysis was carried out using the tool Analyze it (v5.10.9) in Microsoft Excell (2010). Graphs for assay validation were made with GraphPad Prism (7.03).

5. Results & Discussion

5.1 Detection of Underivatized DHT

Initially, we developed a LC-method for the detection of underivatized DHT. DHT elutes at a relatively high percentage methanol, further referred as solvent B. This was determined by the analysis of DHT and DHT-d3 standards. Therefore, a gradient was used starting at 60% solvent B, with a flow of 0.400

mL/min using gradient 1 (Table 2).

DHT and DHT-d3 standards were both extracted with the previously described underivatized method.

After analysing DHT standards, pooled serum samples were used. These samples contained left-over serum from randomly chosen anonymous patients. As displayed in Figure 6, DHT is hardly detected. Almost no peak was shown for DHT around tR 1.78. However, more peaks were observed in the DHT

chromatogram, since serum contains lots of other apolar compounds with similar ion-transitions as DHT. Nonetheless, these peaks do not interfere with the analyte peak.

Underivatized DHT is hard to detect, since DHT only contains neutral functional groups. We expected that a derivatization reagent improves the ionization efficiency, because hydroxylamine and OTB contain a nitrogen atom that can accept a proton [29].

(13)

Figure 6 – (cropped) XIC’s of underivatized DHT and DHT-d3 from pooled serum sample

5.2 Detection of Hydroxylamine Derivative

First, we used hydroxylamine as derivatization reagent. Gradient 2 (Table 2) was used for the detection of the DHT-oxime of hydroxylamine. Hydroxylamine can be considered as either an oxide of ammonia (NH3O) or a hydroxyl derivative (NH2OH) due to the following reaction [30]:

The equilibrium of this reaction lies on the left side (NH2OH), although both forms of the derivative

are known [30]. The DHT-oxime has two isomer forms, a syn- and an anti-isomer (Figure 7), which can cause peak splitting [31]. This phenomenon is commonly observed in literature using hydroxyl-amine as a derivatization reagent [25,28,32].

Figure 7 - DHT-oxime of hydroxylamine anti-isomer (~60%) and syn-isomers (~40%) [33]

As shown in Figure 8, split peaks were observed measuring DHT in pooled serum. Peak splitting is not desirable, as it is unsure whether it is one distorted peak or two compounds that were partially resolved. In addition, it is best to criticize one individual peak, because split peak can vary in ratio. It is known that both peaks were products of the reaction with hydroxylamine, since measurement of DHT standards (1 nmol/L) resulted in the same split peak. We adjusted the gradient to provide separation of the peaks. Using smaller particles can improve peak capacity, deriving from the Van Deemter equation [34]. Therefore, this method uses a C18 column with 1.7 µm particles to achieve a higher resolution. In addition, a flow rate of 0.600 mL/min resulted in better peak shapes compared to a flow rate of 0.400 mL/min. Nevertheless, no base-line separation was achieved. We tried to overcome peak splitting by adjusting the gradient and combining the two peaks as a single peak. Although, this resulted in a non-Gaussian shaped peak. Derivatization with hydroxylamine will result in peak splitting and both peaks were used for quantification. Since two peaks were not completely desirable, another derivatization reagent was used.

+ ₊

DHT-d3

(14)

Figure 8 – (cropped) XIC’s of DHT and DHT-d3-oxime of hydroxylamine (HA) from pooled serum sample using 1.7 µm

particles with a flow of 0.600 mL/min

5.3 Detection of O-tert-butyl hydroxylamine Derivative

Next, derivatization with OTB was evaluated. OTB reacts similar to hydroxylamine at the same keto-side of the DHT-molecule. DHT-OTB contains three extra methyl groups, providing better ionization than the oxime of hydroxylamine. For the detection of DHT-oxime of OTB, gradient 3 was used (Table

2). In contrast to the LC-method used for hydroxylamine, this method uses 2.6 µm particles and a

flow of 0.400 mL/min.

Derivatization with OTB resulted in a single peak compared to the double peak formation seen with hydroxylamine derivatization. Figure 9 displays the chromatograms of DHT-OTB and DHT-d3-OTB

extracted from serum. Detection of DHT-OTB resulted in an analyte peak with additional peaks. It was unsure if these interfering peaks were a result of other endogenous compounds present in serum. Therefore, we performed an experiment using double stripped FBS. Figure 10 illustrates that charcoal stripped FBS was cleaned from steroids. We compared blanks of FBS and double stripped FBS, where no peaks were observed after double charcoal stripping (Figure 10b). This means that the possible interfering peaks were a result of other apolar compounds present in human serum with similar ion transitions as DHT-OTB. It was unsure what compounds caused the interfering peaks. Yue

et al. 2012 [23] used a LC-method with similar OTB derivatization, also resulting in additional peaks

near the analyte peak. However, this method used a two-dimensional LC separation. This means that the possible interfering compound(s) were hard to separate from the analyte, because two-dimensional LC improves selectivity [35]. Nevertheless, the analyte peak can be distinguished from the other peaks and used for quantification.

DHT-d3-HA m/z 309.291 >77.00

DHT-HA

m/z 306.256 > 105.200

(15)

Figure 9 – (cropped) XIC of DHT and DHT-d3-oxime of OTB from pooled serum using 2.6 µm particles with a flow of 0.400

mL/min

Figure 10 – (cropped) XIC’s of (a) FBS blank (b) double stripped FBS blank

We developed three LC-methods for the detection of DHT. A method comparison was performed comparing the three methods: underivatized, derivatization with hydroxylamine, and derivatization with OTB. With this method comparison we determined which method provided the best detection of DHT in serum. a) b) DHT-d3-OTB m/z 365.016 > 309.100 DHT-OTB m/z 362.030> 306.300 DHT-OTB m/z 362.030> 306.300 DHT-OTB m/z 362.030> 306.300

(16)

5.4 Method Comparison

Subsequently, a method comparison was performed by comparison of the underivatized method and the two derivatization methods. A Passing-Bablok regression analysis was used, since we expected that all three methods should result in the same measurement of DHT levels. For Passing-Bablok regression analysis, median calculated DHT concentrations of all three methods were used, since there was no reference method for comparison. To assess the agreement of each method, the Pearson correlation coefficient was calculated.

In this experiment, 24 serum samples containing high Tsto concentrations were obtained from left-over serum. All samples were anonymized, and DHT concentrations were measured in duplicate with all three methods. The hydroxylamine method resulted in peak splitting. To determine if differences are observed by integration, the left peak, right peak, and both peaks were used for quantification. As illustrated in Figure 11, DHT concentrations of each method were plotted against median DHT concentrations. Measurements with the underivatized method resulted in quantification of only 12 samples. The other samples were not detectable with this method. The two derivatization methods were able to detect and quantify (S/N > 10) all 24 serum samples.

The underivatized method showed the lowest Pearson-correlation (r = 0.983). The correlation of DHT measured with the hydroxylamine method fitted best to the linear regression with a Pearson-correlation of 0.996, and resulted in a slope of 0.9705. Therefore, the hydroxylamine method gave the best linearity. Measurements performed with the OTB method detected similar DHT levels to the hydroxylamine method.

For method imprecision (CV%) calculation, duplicate measurements were performed in one run. The CV was calculated with Formula 1,

𝐶𝑉 = ( 𝑆𝐷

𝑚𝑒𝑎𝑛) ∗ 100% ,

Formula 1 – Calculation of method imprecision (CV%)

where the standard deviation (SD) is divided by the mean of the duplicates. CV’s for method underivatized, hydroxylamine, and OTB were 20.2%, 5.1%, and 7.9%, respectively. The acceptance criterion for method imprecision was ≤ 20%, both hydroxylamine and OTB method meet this requirement. The underivatized method gave the highest CV, because this method showed most variation between the duplicates. This variation was probably caused by lack of sensitivity, because DHT is hard to ionize due to its stable structure.

Next, we investigated if all methods were able to quantify relatively low DHT levels. To gain insight into these DHT levels, we obtained serum from male patients treated with abiraterone or enzalutamide. In addition, serum from pre- and post-menopausal women were included into this experiment. These measurements were performed in duplicate with all three methods, and the results are listed in Table 4.

(17)

Underivatized n = 12

Figure 11 – Passing-Bablok regression and Pearson correlation coefficient (r) for DHT (n = number of duplicates; CI = confidence interval; underiv = underivatized; HA = hydroxylamine; OTB = O-tert-butyl hydroxylamine)

Table 4 – DHT concentrations in nmol/L from male patients treated with abiraterone or enzalutamide, and pre- and post-menopausal women, measured with methods underivatized, hydroxylamine (HA), and O-tert-butyl hydroxylamine (OTB)

(ND = not detected, SD = standard deviation)

Underiv (nmol/L) HA left (nmol/L) ± SD HA right (nmol/L) ± SD HA both (nmol/L) ± SD OTB (nmol/L) ± SD Abiraterone N/D N/D N/D N/D 0.014 to 0.032 ± 0.009 Enzalutamide N/D N/D N/D N/D 0.014 to 0.112 ± 0.039 Pre-menopausal N/D 0.174 to 0.280 ± 0.045 0.114 to 0.227 ± 0.048 0.147 to 0.258 ± 0.052 0.149 to 0.337 ± 0.071 Post-menopausal N/D N/D N/D N/D 0.042 to 0.138 ± 0.046 Hydroxylamine n = 24 O-tert-butyl Hydroxylamine n = 24 Method Slope (95%-CI) Intercept (95%-CI) r (95%-CI) Underiv 1,062 (0,8943 to 1,275) -0,1875 (-0,4750 to 0,05645) 0,983 HA 0,9705 (0,9225 to 1,049) -0,01829 (-0,8056 to 0,02044) 0,996 OTB 1,074 (1,003 to 1,149) -0,005267 (-0,06013 to 0,04742) 0,994

(18)

The data in Table 4 illustrate that DHT can only be quantified in serum of patients treated with abiraterone or enzalutamide after derivatization with OTB. Reference DHT values are <0.30 nmol/L in hormonal treated patients [3]. A study of Kim et al. 2014 [18] reported DHT levels <0.086 nmol/L of abiraterone-treated patients. We reported DHT levels lower than these reference values with the OTB assay. Furthermore, we were able to measure DHT levels in serum from enzalutamide-treated patients with the OTB assay. DHT levels of enzalutamide-treated patients were reported around 0.011 nmol/L (± 0.005) in a study of Montgomery et al. 2017 [36].

DHT levels vary between pre- and post-menopausal women. This can be confirmed with the data from Table 4 where measured DHT levels were lower in post-menopausal women than in pre-menopausal women. DHT levels in post-pre-menopausal women were only detected after derivatization with OTB. DHT levels were quantified with both derivatization methods in serum from pre-menopausal women. However, differences were observed between integration of the DHT-hydroxylamine peaks (left, right, and both peaks). Therefore, peak splitting is not desirable.

DHT is an interesting target for clinical research through its binding to the AR. Various studies reported LC-MS based assays for the detection of DHT. Studies using no derivatization reagent reported LLOQ’s between 0.14 and 0.25 nmol/L [21,37–39]. Other studies implemented a derivatization step into the sample preparation. These studies published LLOQ’s within a range of 0.02 and 0.25 nmol/L [22,24,31]. In this experiment, we reported DHT levels <0.02 nmol/L in abiraterone- and enzalutamide-treated patients with the OTB-method. Because DHT can be detected in serum from hormonal-treated men using derivatization with OTB, this assay was preferred over the underivatized-, and hydroxylamine method.

5.5 Method Optimization of DHT-O-tert-butyl hydroxylamine 5.5.1 Interference Test

Subsequently, the additional peaks near the analyte were evaluated. DHT was measured in serum, containing possible interfering compounds, such as phospholipids and similar structured steroid hormones. To determine what compound(s) possible interfere(s) with the analyte, we performed an interference test using double stripped FBS. In this test, we spiked three FBS pools containing different DHT concentrations (3, 0.5, and 0.06 nmol/L) with one of the structure analogs. These structure analogs were: epitestosterone (EpiT), Tsto, androstenedione, 17-hydroxy progesterone (17OHP), dehydroepiandrosterone (DHEA), cortisol, estrone (E1), and 17β-estradiol (E2). Spiked concentrations of the structure analogs were based on serum reference values. Recoveries between 90 – 110 % were considered acceptable [40]. High DHT recovery was observed after addition of DHEA (Table 5). DHEA eluted at the same retention time as DHT (Figure 12). DHEA has a similar structure to DHT, but a different product ion (Q3). Fragment ions as a result of OTB derivatization are derived from neutral loss of the O-tert-butyl moiety, meaning derivatization with OTB lacks of specificity [23]. A second transition is needed to improve specificity. Nevertheless, no second intensive product ion was observed after tuning the analyte. Therefore, we aimed to improve the LC-method to adequately separate the two steroids.

(19)

Table 5 – Recovery% DHT from interference test using structure analogs: epitestosterone (EpiT), testosterone (Tsto), androstenedione (Andro), 17-hydroxy progesterone (17OHP), dehydroepiandrosterone (DHEA), cortisol, estrone (E1), and 17β-estradiol (E2) (SD = standard deviation)

Interference DHT Recovery% SD EpiT (10 nmol/L) 99 10,0 Tsto (10 nmol/L) 92 9,7 Andro (8 nmol/L) 99 13,9 17-OHP (3 nmol/L) 105 7,5 DHEA (7 µmol/L) 192 1,2 Cortisol (350 nmol/L) 107 9,9 E1 (0.3 nmol/L) 104 5,0 E2 (0.5 nmol/L) 106 8,8

Figure 12 – (cropped) XIC’s of (a) DHEA- (10 nM) and (b) DHT-oxime (1 nM) of OTB from a serum sample [13]

5.5.2 Optimization of LC-Method

Next, we evaluated gradient 3 (Table 2) used with OTB derivatization to improve separation of DHT-OTB from DHEA. To provide a different separation, a different column can be used. To determine whether DHT-OTB can be separated from DHEA-OTB, a phenyl hexyl column with 1.7 µm particles was used. Accordingly, MS settings were optimized and DHEA was included into the MRM (Table 6). Using the phenyl hexyl column with gradient 3 provided separation between DHEA (tR = 2.58

minutes) and DHT (tR = 2.73 minutes) (Figure 13).

m/z 359.930 > 304.200 a)

m/z 362.030 > 306.300 b)

DHEA

(20)

Figure 13 – (cropped) XIC’s of separated DHEA- (10 nM) and DHT-oxime (1 nM) of OTB

Although, we observed interferences between DHT, EpiT and Tsto using the phenyl hexyl column (Figure 14). This interference was not seen using the C18 column, where Tsto and EpiT were separated from DHT. Tsto and EpiT both have a similar molecular structure to DHT (Figure 14a+b). EpiT and Tsto only differ in configuration, which makes EpiT a diastereomer of Tsto, and both steroids show split peaks in the chromatogram [41]. To monitor Tsto and EpiT, both steroids were tuned and the optimal ionization settings were included into the MRM settings (Table 6).

Table 6 – MRM settings for the internal standard-, dihydrotestosterone- (DHT)-, dehydroepiandrosterone- (DHEA), epitestosterone (EpiT), and testosterone- (Tsto) oxime of O-tert-butyl hydroxylamine (OTB) (DP = declustering potential, EP = entrance potential, CE = collision energy, CXP = collision cell exit potential, Da = Dalton; msec = mili-seconds; V = volt) Compound Name Q1 Mass (Da) Q3 Mass (Da) Dwell Time (msec) DP (V) EP (V) CE (V) CXP (V) DHT-d3-OTB 365.016 309.100 60.0 91.000 5.000 40.000 12.000 DHT-OTB 362.030 306.300 60.0 111.000 5.000 25.000 5.000 DHEA-OTB 359.930 304.200 60.0 46.000 10.000 25.000 16.000 EpiT-OTB 360.017 304.100 60.0 116.000 10.000 31.000 42.000 Tsto-OTB 359.900 304.100 60.0 194.000 10.000 11.000 22.000 DHEA DHT Time, min

(21)

Figure 14 – (cropped) XIC’s of (a) Epitestosterone- (1 nM), (b) testosterone- (1 nM) and (c) DHT-oxime (1 nM) of OTB m/z 360.017 > 304.100 a) m/z 362.030 > 306.300 c) Time, min Time, min Time, min m/z 359.900 > 304.100 b) 2.77 2.82 DHT Tsto EpiT

(22)

Subsequently, we further optimized chromatographic separation, which lead to establishment of gradient 4 (Table 2). The gradient was optimized by using a less steeper increase of solvent B, providing better separation between DHT, Tsto, and EpiT. A linear gradient was used at a percentage of 80% solvent B, to elute the compounds. With this gradient, DHEA was more separated from DHT compared with gradient 3 (Figure 15a). In addition, the separation between DHT and Tsto has been improved (Figure 15b). Nonetheless, EpiT was still not separated from DHT using gradient 4 (Figure

15c). We performed an interference test with charcoal stripped FBS using this gradient method, and

DHT recoveries were calculated (Table 7). This interference test included three serum pools of high, medium, and low DHT levels.

Testosterone DHT DHT DHEA a) b) Time, min Time, min Continuing Figure

(23)

Figure 15 – (cropped) XIC of (a) DHEA- (10 nM) and DHT-oxime (1 nM) of OTB (b) Testosterone- (1 nM) and DHT-oxime (1 nM) of OTB (c) EpiT- (1 nM) and DHT-oxime (1 nM) of OTB with optimized gradient method

Table 7 – Recovery% DHT from interference test with epitestosterone (EpiT), Testosterone (Tsto), and dehydro-epiandrosterone (DHEA) (SD = standard deviation)

DHT Recovery% ± SD

Interfering Steroid Pool 1 (DHT = 3 nmol/L) Pool 2 (DHT = 0.5 nmol/L) Pool 3 (DHT = 0.06 nmol/L Tsto (1 nmol/L) 103 ± 3.0 99 ± 4.8 99 ± 6.6 DHEA (10 nmol/L) 103 ± 2.6 102 ± 3.9 97 ± 7.6 EpiT (1 nmol/L) 102 ± 4.0 96 ± 0.7 202 ± 38.1 EpiT (500 pmol/L) 103 ± 0.53 105 ± 1.8 136 ± 10.6 EpiT (100 pmol/L) 99 ± 3.9 105 ± 0.9 109 ± 1.7

The recoveries in Table 7 show no interferences of Tsto or DHEA in all three pools, meaning these steroids do not affect calculated DHT concentrations. EpiT was not separated from DHT. Nevertheless, 1 nmol/L of EpiT interfered only in pool 3 containing low DHT concentrations. Therefore, we tested different EpiT concentrations (500 and 100 pmol/L) to determine if these concentrations resulted in acceptable DHT recoveries. We observed acceptable DHT recoveries and no interfering peaks at EpiT concentrations below 100 pmol/L.

The biosynthesis of EpiT only occurs in the testis, and is excreted in 3% of Tsto levels. However, Tsto is not converted into EpiT [42,43]. EpiT levels are highest during puberty and decrease by age. Mostly, EpiT levels in prostatic tissue were reported in literature and detected around 2.96 nmol/L in men with prostate cancer [44,45]. Nevertheless, no correlation was found between EpiT levels in prostatic tissue and serum. EpiT is frequently detected in serum with doping-control tests in sport, but rarely in healthy men and women [42]. Use of exogenous Tsto can relatively increase the amount

c)

DHT

EpiT

(24)

of Tsto. Consequently, ratios between Tsto and EpiT change, and are used as a measurement scale for doping-control tests [42].

Probably, because EpiT is hardly detected in serum, studies on EpiT were rare [43]. Therefore, no reference values were available. To estimate EpiT levels in serum, we evaluated samples measured with a steroid LC-MS/MS assay (Antoni van Leeuwenhoek, Amsterdam, The Netherlands). Charcoal stripped FBS was spiked with 100 pmol/L EpiT and measured the steroid assay. With this assay, Tsto levels were analyzed in serum of healthy men and women, and hormonal treated men. When EpiT was detected, we compared the peak area to the corresponding Tsto peak. In addition, we compared the EpiT peak with the measured peak area of 100 pmol/L EpiT. The results are listed in Table 8. For healthy men, EpiT was detected in 15 of 96 serum samples with Tsto concentrations between 3.6 and 24.3 nmol/L. Detected EpiT peaks were around 0.57% of the corresponding Tsto peak. These EpiT peaks were around 67% (~67 pmol/L) of the measured peak area of 100 pmol/L EpiT. Only one EpiT peak resulted in an area higher than peak areas for 100 pmol/L EpiT. This sample contained a relatively high Tsto concentration of 16.4 nmol/L. Within this healthy population, no EpiT peaks were observed in women below 50 years. For women above 50 years, one EpiT peak was detected with a similar peak area to 100 pmol/L EpiT, indicating that interference of EpiT is frequently low in serum measured from healthy women. Serum measured from men treated with hormonal therapy with Tsto concentrations lower than 3 nmol/L, showed no EpiT peaks higher than 100 pmol/L EpiT peaks. Decrease of Tsto concentrations as a result of hormonal therapy, result in decreased DHT levels. Therefore, interference of EpiT will be negligible at low DHT levels. In serum measured from men containing Tsto concentrations higher than 3 nmol/L after hormonal treatment, EpiT was detected in 60 samples, containing only 4 EpiT with higher peak areas than 100 pmol/L EpiT.

The data display appearance of EpiT in only 16 of 291 measured serum from healthy men and women, and 65 of 763 measured serum from hormonal treated men. This illustrates that EpiT was detected in 5.5% of serum from the healthy population, where only 0.3% from the healthy population an EpiT peak was quantified as >100 pmol/L. An EpiT peak was observed in 8.5% of serum from men treated with hormonal therapy, where only 0.5% was quantified as >100 pmol/L.

Table 8 – Epitestosterone (EpiT) and testosterone (Tsto) levels measured in serum of healthy and castrated men (Cas) with high and low testosterone levels (SD = standard deviation; n = number of measurements/observations)

Population n [Tsto] in nmol/L ± SD % area EpiT of area Tsto ± SD %EpiT of 100 pmol/L area ± SD n EpiT area >100 pmol/L ± SD n EpiT peak Healthy men (19 to 79 years) 96 12.6 ± 4.2 (3.6 to 24.3) 0.57 ± 0.25 (0.30 to 1.0) (26.0 to 120) 67 ± 25.7 1 15 Healthy women (< 50 years) 90 0.75 ± 0.4 (0.3 to 2.9) - - 0 0 Healthy women (≥ 50 years) 105 0.58 ± 0.29 (0.1 to 1.8) 0.73 100 0 1 Hormonal treated men Tsto < 3 nmol/L 600 0.32 ± 0.15 (0.18 to 0.53) (2.7 to 5.0) 3.7 ± 1.1 (8.0 to 21.0) 13 ± 5.2 0 5 Hormonal treated men Tsto 163 13 ± 5.3 (5.9 to 38.7) 0.45 ± 16 (0.24 to 1.05) 44 ± 30 (6.0 to 147) 4 60

(25)

EpiT was mostly detected in serum containing relatively high concentrations of Tsto. However, the results illustrate that only 1% (1 of 96) of the measurements displayed possible interference of EpiT (>100 pmol/L EpiT) in this population of healthy men. In addition, the EpiT peak was detected in serum containing a relatively high concentration of Tsto (16.4 nmol/L). Therefore, the relationship between DHT levels with corresponding Tsto levels was evaluated in left-over serum containing relatively high Tsto levels. Median DHT concentrations were around 7% (± 5.5) of the Tsto concentration. As displayed in Figure 16, increasing Tsto levels resulted in higher DHT concentrations. However, almost no correlation between Tsto and DHT levels was observed within this population, since R2_{=0,2825. In contrast, it is known that decreased testosterone levels resulting from hormonal}

therapy, show also decreased DHT levels [17–19]. EpiT was mostly detected with concentrations higher than 100 pmol/L in serum containing relatively high (≥ 3 nmol/L) Tsto levels. Therefore, EpiT will probably be infrequently measured in concentration >100 pmol/L in serum containing relatively low DHT levels, and therefore not affect measured DHT concentrations.

Figure 16 – Correlation between measured testosterone and median DHT levels (n=24); y = 0,0418x + 0,3835; R² = 0,2825

5.6. Validation

For the validation of the method using OTB derivatization, analyzes were performed in duplicates including blanks, double blanks and quality controls (QCs).

5.6.1 Calibration Curve

A calibration curve was included in all measurements to monitor mean ratios of DHT standards and calculate DHT concentrations [27]. DHT standards were prepared in 0.9%-NaCl with the concentrations described in Table 1. In addition, we prepared calibrators in single, double, triple, and quadruple charcoal stripped FBS, to compare with 0.9%-NaCl. No differences were observed between double stripped FBS and triple- and quadruple stripped FBS. Calibrators prepared in double stripped FBS and 0.9%-NaCl resulted in similar slopes: 3.50098 and 3.48370, respectively. Preparation

(26)

of the calibrators was preferred in 0.9%-NaCl, because charcoal stripping of FBS is more time consuming. A weighting factor of 1/x2_{was used, because DHT is measured in low concentrations,}

meaning accuracy in low concentrations is more important. This weighting factor lowers the overall error in the methods. The OTB method showed linearity (R2_>0.99).

5.6.2 Stability

Analyte stability must be ensured during sample collection and storage to obtain reliable analytical data. For stability measurements, we distributed serum from three patients within 24 hours after collection, and stored (1) – at room temperature (20°C) and analyzed after 4 and 7 days (Table 9), (2) – at 4°C and analyzed after 7, 14 and 21 days (Table 10), and (3) – at -20°C and analyzed after 21 days and 2 months (Table 11).

DHT stored at room temperature showed relatively low recovery after 4 and 7 days. This degradation can be caused by temperature or exposure to light. Acceptable recoveries were observed after storage at 4°C and -20°C, meaning storage at low temperatures increases stability by delaying degradation of the analyte. DHT storage for 2 months at -20°C, resulted in a recovery higher than storage for 21 days. This was probably a result of inter-variation, since these serum samples were measured in different runs. Storage of DHT is preferred at low temperatures and rooms without light, preventing degradation of the analyte. Degradation of the analyte can result in measurement imprecisions [27].

5.6.3 Carry-Over, Matrix Effects and Recovery

To determine whether this assay suffers from signal carry-over, three blanks and three high calibrators (10 nmol/L) were analyzed. The acceptance criterion for carry-over was <0,1%. No peaks were observed in the blanks, which means no carry-over was determined.

In addition, we determined the total recovery and matrix effect recovery by addition of an academic standard in double charcoal stripped FBS. This resulted in a recovery of 84.3% (post-extraction recovery) including matrix effects, and an total recovery (pre-extraction recovery) of 55.0% (Table

12). This gives an extraction loss of 29.3% during the extraction procedure. The total recovery was

Table 9 – Stability DHT at 20°C (SD = standard deviation, CI = confidence interval)

20°C 4 days 7 days

Recovery(%) 86 80

SD 8.6 8.1

CI (95%) 6.9 6.5

4°C 7 days 14 days 21 days

Recovery(%) 101 98 94

SD 8.4 6.9 5.8

CI (95%) 6.7 5.5 4.6

-20°C 21 days 2 months

Recovery(%) 96 110

SD 9.7 9.8

(27)

relatively low, probably because serum is a complex matrix containing other endogenous compounds – For example phospholipid can cause suppression of the analyte, which is then termed as ion-suppression [45]. Another cause of matrix effects can be co-elution of other steroid hormones. Table 12 – DHT recoveries standard addition experiment

Matrix effects (%) Total recovery (%) Extraction loss (%)

84.3 55.0 29.3

5.6.4 Interferences

We performed an interference test during the assay optimization. Three FBS pools containing different DHT concentrations (3, 0.5, and 0.06 nmol/L) were spiked with hemoglobin, bilirubin, intra-lipids, and structure analogs (Table 13). These structure analogs were: EpiT, Tsto, androstenedione, 17OHP, DHEA, cortisol, E1, and E2. The results were previously discussed and presented in Table 5+ 7. Relatively low DHT recoveries were observed after addition of hemoglobin and intra-lipids. Presence of endogenous compounds can cause ion suppression. Blood contains high concentrations of phospholipids. LLE was performed to extract the analyte and possibly remove phospholipids. Solid phase extraction was probably better to use for sample clean up. Nevertheless, solid phase extraction is more time consuming than LLE. However, probably none of the extraction procedures (LLE, solid phase extraction, and protein precipitation) provide enough selectivity to distinguish the analyte from phospholipids [46]. Acceptable DHT recoveries were observed after detection of structure analogs, except for EpiT. EpiT concentrations below 100 pmol/L resulted in acceptable recoveries and did not affect DHT concentrations.

Table 12 – Recovery% DHT from interference test using hemoglobin, bilirubin, intra-lipids, (SD = standard deviation)

Interference DHT Recovery% SD Hemoglobin 1000 µmol/L 82 10,7 Hemoglobin 500 µmol/L 90 8,7 Hemoglobin 250 µmol/L 88 7,1 Bilirubin 20 µmol/L 99 10,1 Bilirubin 10 µmol/L 108 8,7 Bilirubin 5 µmol/L 105 5,1 Intra-lipids 2% 70 5,5 Intra-lipids 1% 77 3,1 Intra-lipids 0.5% 85 3,5 5.6.5 Linearity

To define linearity, two samples with high (QC1) and low (QC3) DHT concentrations were diluted. The results of the measurements are presented in Figure 17. The DHT method with OTB derivatization showed linearity with R2_{=0.9924 between DHT concentrations of 0.05 and 3.44 nmol/L. QC ratios}

increase with 20%, therefore the slope should have a value around 20 to define linearity. The slope was 19.33 ± 1.2, which showed an acceptable linearity.

(28)

0 2 0 4 0 6 0 8 0 1 0 0 L in e a r it y D H T R a tio Q C 1 :Q C 3 % R e la ti v e t o Q C -r a n g e 1:4 2:3 3:2 4:1

Figure 17 – Linearity of DHT method with OTB derivatization (y = 19.33x+3.875); number of values per point = 4

5.6.6 Reproducibility and Lower Limit of Quantification (LLOQ)

The reproducibility was determined according to CLSI guidelines [27]. According to these guidelines, three reference levels (QC’s) were measured four times for seven runs. These reference values (high (QC1), medium (QC2) and low (QC3)) contained the following DHT concentrations (Table 13): Table 13 – Mean DHT concentrations and method imprecision (CV%) of reproducibility experiment (QC = quality control) (n = number of measurements)

QC 1 QC 2 QC3

n 28 20 28

Mean (nmol/L) 3.35 0.27 0.05

CV% 7 6 8

All QC-levels showed total method imprecision <20%. In QC level 2, higher DHT levels were measured in the first two duplicate measurements, resulting in a relatively high method imprecision. This imprecision can be caused by pipetting errors or different peak integration. No differences were observed for QC levels 1 and 3 in these two runs. Therefore, the first two measurements of QC 2 were excluded. A Levey Jennings graph is used to monitor quality controls of the measurements. The reproducibility was acceptable for all QC-levels.

To determine the lower limit of quantification (LLOQ), we used three pools containing low DHT levels. Criteria for defining the LLOQ were: S/N >10, and CV <20%. The highest calibrator in the calibration curve is used as upper limit of quantification (10 nmol/L).

Table 14 - Mean DHT concentrations and method imprecision (CV%) to determine LLOQ of DHT

Pool 1 Pool 2 Pool 3

Mean (nmol/L) 0.014 0.011 0.009

CV% 11 13 25

S/N 14 12 10

The data in Table 14 illustrate that the LLOQ is 0.014 nmol/L with a S/N >10 and CV <20%. Relatively high method imprecision was observed in pool 2 and 3, since DHT concentrations were lower and resulted in less reproducible measurements. Although, 0.009 nmol/L DHT was also detected with a S/N ≥10.

(29)

We compared method imprecisions and LLOQ’s of DHT detected with other assays. Table 15 displays different DHT-assays using derivatization reagents or no derivatization step. Assays without using a derivatization step reported varying LLOQ’s, namely between 0.143 and 0.250 nmol/L. These assays used different columns, such as C8, C18 or a cyano column. In contrast to other assays, two underivatized methods used APPI or APCI instead of ESI. The assay using APCI as ion source reached a lower LLOQ compared to assays using ESI, probably because APCI provides high frequency collisions to thermalize the analyte and increases fragment ion intensity [47,48]. Derivatization with picolinic acid resulted in similar LLOQ’s as derivatization with hydroxylamine (Kalhorn et al. 2007, Hakkinen et

al. 2019) and OTB (Yue et al. 2012). Methods using derivatization with picolinic acid detected DHT in

serum with a reversed phase C18 column and ESI ionization. The two assays using hydroxylamine showed either a relatively high method imprecision and high LLOQ (Kalhorn et al. 2007), or a relatively low LLOQ with a lower method imprecision (Hakkinen et al. 2019). Both assays used a similar reversed phase C18 column, and a similar extraction procedure. There is one other assay using derivatization with OTB. Nevertheless, this assay used a two-dimensional LC system with a cyano- and a C8 column, and reported a method imprecision of 7% and a LLOQ of 0.0215 nmol/L. However, we reached a lower LLOQ than reported by Yue et al. 2012, using similar derivatization. We reported a method imprecision (mean 7%) similar to other DHT-assays. Although, we detected DHT in serum from abiraterone- and enzalutamide treated patients with a LLOQ of 0.011 nmol/L. Not all assays described in Table 15 detected DHT in serum from hormonal treated patients, because DHT was mostly detected in serum from healthy volunteers.

(30)

Table 15 – Assay comparison LC-MS assays for the detection of DHT (LLOQ = lower limit of quantification, CV% = method imprecision)

Method System Column Ion

Source (mode) Derivatization Mean CV(%) LLOQ (nmol/L) Kalhorn et al. (2007)

Waters Aquity UPLC, Micromass Premiere-XE tandem quadrupole MS Waters BEH C18 column (2.1 x 100 mm, 1.7 µm) ESI (+) Hydroxylamine 10 0.250 Shiraishi et al. (2008) Shimadzu SCL-10Avp system, API 5000 LC-MS/MS Thermo Hypersil GOLD C18 column (100 mm x 1 mm, 3 µm) ESI (+) No 5 0.173 Yamashita et al. (2009) Agilent 1100, API-4000 triple stage quadrupole MS

Cadenza CD-C18 (150 mm × 3 mm, 3 µm)

ESI (+) Picolinic acid 8 0.0215

Harwood et al. (2009)

Shimadzu Prominence, API-5000 triple-quadrupole MS Supelcosil LC-8-DB (7.5 cm × 3 mm, 3 μm) APPI (+) No 4 0.215 Ghoshal et al. (2010)

Cohesive Technologies Aria TLX-1 HTLC, Finnigan TSQ Quantum Ultra MS/MS Thermo Scientific, BetaBasic CN-column (50 x 2.1 mm, 5 μm) APCI (+) No 7.5 0.143 Yue et al. (2012) 2D LC system, API-5000 triple quadrupole MS 1_{D: Zorbax Eclipes} XDB-CN, 12.5x2.1 mm, 5 µm 2_{D: Luna C8, 3 µm,} 100 A, 100x2 mm ESI (+) O-tert-butyl hydroxylamine 7 0.0215 Owen et al. (2016)

Waters Acquity UPLC HSS C18 SB column (2.1x 50 mm, 1.8 µm) ESI (+) No 8 0.250 Gorityala et al. (2018) LC-20AD HPLC Shimadzu, triple quadrupole MS/MS C18 column (2.1 mm × 100 mm, 3.5 μm)

ESI (+) Picolinic acid 8 0.215

Hakkinen et al. (2019)

Agilent 1290 Rapid Resolution, Agilent 6495 tripe stage quadrupole MS

Acquity UPLC CSH C18 (2.1 x 100 mm, 1.7 μm)

(31)

7. Conclusion

Three different DHT-assays were developed by LC-MS/MS and compared. The DHT-assay combined with OTB derivatization was preferred, since this assay was able to detect DHT in serum from hormonal treated patients, and pre- and post-menopausal women. This assay enabled to detect DHT from serum with a LLOQ of 0.011 nmol/L. The method used LLE to extract the analyte from serum, and a reversed phase phenyl hexyl column for detection. With this assay we were able to separate DHT from interfering compounds. In addition, possible interference were negligible in serum from women and hormonal treated male patients, and did not affect DHT concentrations. For healthy men, interferences were infrequently observed. Despite the fact that reference DHT levels are currently poorly understood, this assay can be adopted for measurements of DHT in serum from healthy- and diseased men and women. For further clinical research involving prostate cancer patients, reference values have to be evaluated to compare DHT concentrations of males treated with hormonal therapy with clinical values related to prostate cancer. Hormonal treatments – abiraterone or enzalutamide – compete with Tsto at the AR in the prostate cell. This results in decreased Tsto production and lower DHT production. To define the possible relationship between DHT levels in the prostate and serum DHT levels, it is important to monitor changes of serum DHT levels before and after hormonal treatment. Consequently, this assay can be helpful in understanding the serum biological responses of patients with prostate cancer.

(32)

8. References

[1] K. Hyun-kyung et al., “Loss of an Androgen-Inactivating and Isoform- Specific HSD17B4 Splice Form Enables Emergence of Castration-Resistant Prostate Cancer,” Cell Rep., vol. 22, pp. 809–819, 2018. [2] F. Bray, J. Ferlay, I. Soerjomataram, R. L. Siegel, L. A. Torre, and A. Jemal, “Global Cancer Statistics 2018:

GLOBOCAN Estimates of Incidence and Mortality Worldwide for 36 Cancers in 185 Countries,” CA

Cancer, vol. 68, no. 6, pp. 394–424, 2018.

[3] J. H. Gido Snaterse, Jenny A. Visser, Wiebke Arit, “Circulating Steroid Hormone Variations Throughout Different Stages of Prostate Cancer,” Soc. Endocrinol., vol. 24, pp. 403–420, 2017.

[4] T. M. van der Sluis et al., “Intraprostatic testosterone and dihydrotestosterone. Part II: concentrations after androgen hormonal manipulation in men with benign prostatic hyperplasia and prostate cancer,”

BJU Int., vol. 109, pp. 183–188, 2011.

[5] T. Cook and W. P. Sheridan, “Development of GnRH antagonist for prostate cancer: New approaches to

treatment,” Oncologist, vol. 5, pp. 162–168, 2000.

[6] D. Robinson et al., “Androgen deprivaton therapy for prostate cancer and risk of dementia,” BJU Int., 2019.

[7] M. A. Titus, M. J. Schell, F. B. Lih, K. B. Tomer, and J. L. Mohler, “Testosterone and Dihydrotestosterone Tissue Levels in Recurrent Prostate Cancer,” Clin. Cancer Res., vol. 11, no. 13, pp. 4653–4657, 2005.

[8] N. Ammannagari and S. George, “Anti-Androgen Therapies for Prostate Cancer,” Am. J. Hematol.

Oncol., vol. 11, no. 2, pp. 15–19, 2015.

[9] J. A. Locke et al., “Androgen Levels Increase by Intratumoral De novo Steroidogenesis during

Progression of Castration-Resistant Prostate Cancer,” Am. Assoc. Cancer Res., vol. 68, no. 15, pp. 6407– 6415, 2008.

[10] L. . M. Walter and C. E. Flück, “Chapter 13 - Adrenal cortex and its disorders,” in Pediatric

Endorcrinology, 2014, pp. 471–532.

[11] P. Parth K. Modi MD, Izak Faiena MD, Isaac Yi Kim MD, “Prostate Cancer (Second Edition),” in Science

and Clinical Practice, 2016, pp. 21–28.

[12] S. Xu et al., “The inhibitory effects of AR/miR-190a/YB-1 negative feedback loop on prostate cancer and underlying mechanism,” Sci. Rep., vol. 5, 2015.

[13] H. H. van Rossum, A. M. Bergman, and E. Lentjes, “Analytical Challenges and Potential Applications of Sex Steroid Hormone Analysis in Breast and Prostate Cancer Patients,” Chromatographia, vol. 78, no. 5–6. pp. 359–365, 2014.

[14] V. A. Randall, “Role of 5α-reductase in Health and Disease,” Bailliere Tindall, vol. 8, no. 2, pp. 405–431, 1994.

[15] J. L. Mohler and Titus, “Tissue Levels of Androgens in Castration-Recurrent Prostate Cancer,” in

Androgen Action in Prostate Cancer, D. Tindall and J. Mohler, Eds. Springer, 2009, pp. 553–565.

[16] B. J. Feldman and D. Feldman, “The Development of Androgen-Independent Prostate Cancer,”

Macmillan Mag. Ltd, vol. 1, pp. 34–45, 2001.

[17] R. S. Swerdloff, R. E. Dudley, S. T. Page, C. Wang, and W. A. Salameh, “Dihydrotestosterone:

Biochemistry, physiology, and clinical implications of elevated blood levels,” Endocr. Rev., vol. 38, no. 3, pp. 220–254, 2017.

[18] W. Kim et al., “Sequential use of the androgen synthesis inhibitors ketoconazole and abiraterone acetate in castration-resistant prostate cancer and the predictive value of circulating androgens,” Clin.

Cancer Res., vol. 20, no. 24, pp. 6269–6276, 2014.

[19] M. S. Rothman et al., “Reexamination of Testosterone, Dihydrotestosterone, Estradiol and Estrone levels across the Menstrual Cycle and in Postmenopausal Women Measured by Liquid Chromatography Tandem Mass Spectrometry,” vol. 76, no. 720, pp. 177–182, 2012.

[20] K. Yamashita et al., “Development of highly sensitive quantification method for testosterone and dihydrotestosterone in human serum and prostate tissue by liquid chromatography–electrospray ionization tandem mass spectrometry,” Elsevier, no. Steroids 74, pp. 920–926, 2009.

[21] C. Wang et al., “Validation of a testosterone and dihydrotestosterone liquid chromatography tandem mass spectrometry assay,” Steroids, vol. 73, no. 13, pp. 1345–1352, 2008.

[22] L. Owen, F. Wu, R. Buttler, and B. Keevil, “A direct assay for the routine measurement of testosterone, androstenedione, dihydrotestosterone and dehydroepiandrosterone by LC-MS/MS,” Ann. Clin.

Biochem., 2015.

[23] B. Yue, A. L. Rockwood, and M. M. Kushnir, “Enhanced sensitivity for analysis of carbonyl containing compounds using mass spectrometry,” US 2012/0126107 A1, 2012.