Salivary Areca and tobacco alkaloids for bioverification in the Betel Nut Intervention Trial (BENIT)

Abstract

Background:

The Betel Nut Intervention Trial (BENIT) is the first known randomized controlled intervention trial designed to help minority Pacific Islanders in Guam and Saipan quit chewing the carcinogenic Areca catechu nut (AN). We report the BENIT’s saliva bioverification results against the self-reported chewing status (‘quitter’ or ‘chewer’) at day 22 follow-up.

Material and Methods:

AN-specific (arecoline, arecaidine, guvacoline, guvacine) and tobacco-specific (nicotine, cotinine, hydroxycotinine) alkaloids were analyzed in saliva from 176 BENIT participants by an established and sensitive liquid chromatography mass spectrometry-based assay.

Results:

The combined four AN alkaloid levels decreased from baseline in quitters (n=50) and chewers (n=108) by 32% and 9%, respectively. In quitters, decreases were significant for arecoline (p=0.044)—the most prominent AN alkaloid, along with arecaidine (p=0.042) and nicotine (p=0.011). In chewers, decreases were significant only for hydroxycotinine (p=0.004). Similar results were obtained when quitters and chewers were stratified by treatment arm.

Discussion:

Salivary AN alkaloid levels generally agreed with self-reported chewing status, which suggests the former can be used to verify the latter.

Conclusion:

Our results can help to objectively evaluate compliance and program effectiveness in AN cessation programs.

Graphical Abstract

In an Areca nut cessation trial, the combined levels of four Areca alkaloid (Arecoline, Arecaidine, Guvacoline, Guvacine) decreased from baseline in quitters (n=50) and chewers (n=108) by 32% and 9%, respectively. In quitters, decreases were significant for arecoline (p=0.044)—the most prominent AN alkaloid, along with arecaidine (p=0.042) and nicotine (p=0.011). In chewers, decreases were significant only for hydroxycotinine (p=0.004).

Introduction

The endosperm from the drupe fruit of the oriental palm Areca catechu Linn, colloquially referred to as betel nut or Areca nut (AN) [1], is chewed by approximately 600 million people worldwide [2]. In the U.S. territory of Guam, 11% of the adult population chew AN [3] whereas the chewing prevalence in the U.S. Commonwealth of Saipan is estimated at 24% [4]. In these areas, the majority of chewers, denoted ‘Class 2’ chewers, consume AN as a betel quid (BQ) consisting of the AN wrapped in a Piper betle leaf with tobacco (mostly from cigarettes) and slaked lime (calcium hydroxide) [5]. Class 2 chewers are at greater risk than ‘Class 1’ chewers (who chew AN alone or sometimes with slacked lime) for health problems attributable to AN/BQ use [6] including leukoplakia, oral submucous [7, 8], and particularly oral squamous cell carcinoma (OSCC), the most common malignancy of the head and neck region [9, 10] with one of the highest mortality rates among all cancers [11].

Despite the global prominence of AN/BQ chewing and its associated heath burdens, there is no established framework for its control, and prevention and cessation attempts based on evidence-based practices have only begun to develop [12]. For that reason, we recently implemented the Betel Nut Intervention Trial (“BENIT”; registered at ClinicalTrials.gov, trial #NCT02942745), the first known randomized controlled intervention trial specifically designed for AN cessation in Guam and Saipan in the Mariana Islands. The BENIT aims to address the disproportionally high extent of AN/BQ related health disparities, including oral cancer, among minority Pacific Islanders [13] compared to the U.S. population [13]. During the BENIT, participants donated saliva samples and completed questionnaires with self-reported AN/BQ consumption used to designate them either as former chewers, which we refer to as ‘quitters’, or as non-quitters/continuing chewers, which we refer to as ‘chewers.’

Self-reports are the most commonly used methods to collect exposure information and assess quitter prevalence in AN-based studies [14–18]. However, self-reported data are subject to misreporting due to many reasons including recall and social desirability bias and can, therefore, sometimes be unreliable [19]. Biomarkers, on the other hand, are objective indicators that can be reliably measured from various biological specimens, including saliva. We and others have shown that arecoline, arecaidine, guvacoline, and guvacine are specific to AN chewing [1, 9, 20, 21] and can be reliably measured from human saliva and other biological samples [20]. Therefore, these Areca alkaloids can serve as ideal biomarkers for AN exposure to verify self-reported AN use. To our knowledge, no study has used salivary biomarkers to verify self-reported data from participants in an AN/BQ cessation trial.

The aim of this study was to analyze the predominant AN-related (arecoline, arecaidine, guvacoline, and guvacine; Figure 1) and tobacco-related (nicotine, cotinine, and hydroxycotinine; Figure 1) alkaloids in saliva samples collected from BENIT participants at baseline (BL) and day 22 (D22; 1-week following the end of active intervention), then compare the measured levels to the self-reported status of quitters and chewers.

Figure 1.

Chemical structures of Areca nut and Tobacco-specific alkaloids

Method

The BENIT

The study design of the BENIT (registered at ClinicalTrials.gov - NCT02942745) was described in detail previously [22]. Briefly, study participants in Guam and Saipan were randomized into intervention condition or control condition groups. Class 2 chewers were the target population for BENIT because these chewers are the predominant class of chewers in Guam and Saipan, and have far greater detrimental health risks than Class 1 chewers [6]. Participants randomized to the intervention group received an intensive five-session, in-person intervention that was administered over a 22-day period. Intervention participants also received a booklet containing advice on how to quit chewing AN/BQ. The booklet, entitled “Quitting Betel Nut,” included information about the health risks associated with AN/BQ chewing as well as advice and strategies for quitting and maintaining AN/BQ abstinence. Participants randomized to the control group received exclusively the “Quitting Betel Nut” booklet. All participants completed survey assessments at BL, D22, and month 6 (M6), donated ca. 1–2 mL saliva during those periods, and were made aware that their saliva samples would be bioverified against their survey answers. Saliva samples were collected via “passive drool” into 20 mL conical polypropylene tubes then stored at −20°C until shipment to the University of Hawaii (UH) Cancer Center on dry ice, where samples were stored at −80°C until analysis by a sensitive and validated liquid chromatography mass spectrometry (LCMS) assay [20, 21]. All participants were compensated for their time and all received the same compensation.

BENIT inclusion criteria were as follows: 1) self-identified BQ chewer for at least one year with a minimum thrice weekly chewing frequency; 2) 18 years of age or older; 3) willingness to attempt to quit chewing BQ during the intervention; 4) willingness to participate in the five, in-person intervention sessions to take place over approximately 22 days; and 5) English literacy. Pregnant women were excluded from participation. This study was approved by the University of Guam (UOG). The UH ceded authority to UOG for this study (UH CHS #24078). All participants signed an approved written consent form.

Chemicals

Ammonium hydroxide (NH₄OH), Arecoline hydrobromide, guvacine hydrochloride, nicotine, cotinine, hydroxycotinine, nicotine-d₄, and cotinine-d₃ were purchased from Sigma-Aldrich (St Louis, MO, USA). Guvacoline hydrobromide, Arecaidine hydrobromide, Arecodine-d₅, and hydrobromide salt were purchased from Medical Isotopes (Pelham, NH). Methanol (MeOH) was purchased from Fisher Scientific (Waltham, MA, USA). All solvents were of LCMS grade.

Standards

An 8-point calibration curve (0.1–5000 ng/mL) was prepared from a 10 μg/mL stock in 0.1% formic acid in MeOH/water (50/50, v/v). 100 μL of each calibrator was mixed with 10 μL of the internal standard (IS) mixture (1 μg/mL of each arecoline-d₅, nicotine-d₄, and cotinine-d₃).

Sample preparation and analysis

One hundred μL aliquots of thawed saliva were mixed with 10 μL of IS solution and 100 μL of acetonitrile then vortexed at 1,650 rpm for 5 minutes followed by centrifugation at 17,000 × g for 10 minutes. The supernatant was transferred to high performance LC (HPLC) vials and 10 μL were injected for LCMS analysis.

LCMS analysis

LCMS analysis was carried out on a model Accela Ultra HPLC system with a CTDC PAL autosampler coupled to a Q-Exactive orbitrap mass spectrometer (all from Thermo Fisher, San Jose, CA). Saliva extracts were injected onto a Kinetex C18 column (150 × 3 mm, 2.6 μm, Phenomenex, Torrance, CA) with a Phenomenex UHPLC C18 pre-column (3.0 mm i.d.) Gradient elution was performed at a flow rate of 250 μL/minute using 10 mM NH₄OH in water (A) and 10 mM NH₄OH in MeOH (B) as follows: 0–10.0 minutes linear gradient from 65%A to 20%A; 10.1–12.0 minutes hold at 20%A; 12.1–16.0 minutes increase linearly to 65%A. Total HPLC time was 16 minutes. Mass analysis was performed under positive ESI in full scan mode. Quantitation was performed with Xcalibur™ software by extracting the calculated exact masses ± 5ppm as follows: arecoline and arecoline-d₅ at +ESI [M+H]⁺ 156.10191 and 161.13329, arecaidine at +ESI [M+H]⁺ 142.086, guvacoline at +ESI [M+H]⁺ 142.08626, guvacine at +ESI [M+H]⁺ 128.07061, nicotine and nicotine-d₄ at +ESI [M+H]⁺ 163.12297 and 167.14808, cotinine and cotinine-d₃ at +ESI [M+H]⁺ 177.10224 and 180.12107, and hydroxycotinine at +ESI [M+H]⁺ 193.09715.

Samples were analyzed in two batches. For batch one, the lower limit of detection (LLOD; defined as signal-to-noise ratio of 3) was 0.1 ng/mL for all seven analytes. For batch two, the LLODs (ng/mL) were as follows: 0.08 (arecoline), 0.03 (arecaidine), 0.30 (guvacoline), 0.20 (guvacine), 0.10 (nicotine), 0.03 (cotinine), and 0.03 (hydroxycotinine). Samples with concentrations below the LLOD were assigned one-half the LLOD concentration.

Data analysis

For each analyte (and total Areca or total tobacco alkaloids), measured saliva levels were analyzed by two approaches: 1) quitters versus chewers, regardless of treatment group (ie, all control and all intervention participants combined) in order to compare self-reported chewing status with measured alkaloid concentrations, overall and separately for quitters and chewers; and 2) all intervention versus all control participants, regardless of chewing status (ie, all chewers and all quitters combined) in order to examine the intervention effect.

Statistical evaluation

Mixed models were used to estimate means for each analyte at each time point by quit status and quit-intervention status for analysis approach 1 and by intervention status for analysis approach 2, accounting for the repeated measures. The analytes were log-transformed to meet the assumptions of homoscedasticity. Means from the model are presented as geometric means after back-transformation. Contrasts are used to statistically test changes over time within subgroup based on 1-sample t-tests and to statistically compare the differences in the changes over time between groups using 2-sample t-tests. Analyses were performed using Excel (Microsoft, Seattle, WA) and SAS (Sixth Edition, Cary, NC), and a p-value of <0.05 was considered significant. The analysis for D22 included only participants who provided both BL and D22 samples (n=168) and, therefore, had valid BL-D22 change values. A similar analysis for M6 included only participants who provided BL and M6 samples (n=42). However, due to the low numbers for M6 samples, caused partially by the Covid-19 epidemic, we did not consider the BL-M6 findings here. To explore the performance of the biomarker to label quitters and chewers, we proposed a cutpoint for alkaloids (30 pg/mL) as the value that maximizes Youden’s index (sensitivity + specificity – 1) [23]. Logistic regression of self-reported chewing status at D22 on all Areca alkaloids combined was used to determine this cutpoint. The percentage quitting at D22 based on the proposed cutpoint was compared between intervention arms using a chi-square test.

Results

Geometric means of measured Areca and tobacco alkaloid concentrations decreased from BL to D22 (Figure 2) in all quitters (n=49–50) by 23% to 60% with significance for arecoline (−57%; p=0.044)—the most prominent Areca alkaloid, arecaidine (−59%; p=0.042), and the tobacco alkaloid nicotine (−60%; p=0.011). In all chewers (n=107–108), all individual alkaloids also decreased from BL, but to a considerably lesser extent (7% to 41%) and not significantly, except for hydroxycotinine (−41%; p=0.004). When the four measured Areca alkaloids were combined (‘all Areca alkaloids’) and the three measured tobacco alkaloids were combined (‘all tobacco alkaloids’), levels decreased in quitters by 32% and 47% (p=0.007) compared to only by 9% and 16% in chewers, respectively. However, between the quitter and chewer groups, none of the differences in these changes were significantly different (data not shown).

Figure 2.

Absolute numbers are geometric means. % numbers are relative changes to BL. None of the % changes to BL were significantly different between all quitters and all chewers.

*p≤0.05; **p≤0.01; ***p≤0.001

When data in each treatment arm were stratified by quitter status, all measured analyte levels in the intervention group decreased from BL in both the quitters (n=40–41; 16–62%) and the chewers (n=34; 5–47%,) (Table 1a). Data among the control group (Table 1b) showed a similar pattern of change from BL for quitters and chewers. In quitters all alkaloids decreased from BL 2–96% (n=9) while decreases of 2–46% were observed for chewers. Neither quitters nor chewers showed statistically significant differences in any of the analytes individually or when combined together (ie, all Areca alkaloids or all tobacco alkaloids). Similarly, none of the analytes showed any significant differences in the changes from BL to D22 between quitters and chewers.

Table 1a.

Measured salivary alkaloid levels at baseline (BL) and day 22 (D22) from participating BENIT quitters and chewers of the intervention arm.

	Intervention
	Quitters					Chewers
Analyte	n	BL3	D223	% change from BL	p4	n	BL3	D223	% change from BL	p4	p (Quitters vs Chewers change)
Arecoline	41	3,526	1,851	−48%	0.264	34	21,063	18,568	−12%	0.842	0.544
Arecaidine	41	595	392	−34%	0.564	34	13,007	16,722	29%	0.751	0.533
Guvacoline	41	765	639	−16%	0.756	34	7,731	6,446	−17%	0.774	0.998
Guvacine	41	513	356	−31%	0.560	34	4,505	2,381	−47%	0.357	0.772
Nicotine	41	87,329	33,610	−62%	0.053	34	172,615	124,670	−28%	0.546	0.388
Cotinine	40	51,710	37,454	−28%	0.504	34	77,485	73,602	−5%	0.922	0.704
Hydroxycotinine	40	1,433	1,114	−22%	0.700	34	9,056	6,145	−32%	0.584	0.888
All Areca alkaloids1	41	11,648	8,665	−26%	0.566	34	72,152	70,503	−2%	0.967	0.722
All tobacco alkaloids2	40	176,079	89,202	−49%	0.132	34	353,506	258,762	−27%	0.522	0.579

¹arecaidine + guvacine + guvacoline + arecoline

²nicotine + cotinine + hydroxycotinine

³geometric means (pg/mL)

⁴paired t-test n= number of participants

Table 1b.

Measured salivary alkaloid levels at baseline (BL) and day 22 (D22) from participating BENIT quitters and chewers of the intervention arm.

	Control
	Quitters					Chewers
Analyte	n	BL3	D223	% change from BL	p4	n	BL3	D223	% change from BL	p4	p (Quitters vs Chewers change)
Arecoline	9	1,554	274	−82%	0.208	74	19,296	18,475	−4%	0.928	0.246
Arecaidine	9	2,562	113	−96%	0.083	74	12,033	6,516	−46%	0.327	0.188
Guvacoline	9	1,551	190	−88%	0.172	74	4,392	3,699	−16%	0.748	0.236
Guvacine	9	2,964	1,878	−37%	0.774	74	5,225	5,117	−2%	0.970	0.796
Nicotine	9	25,202	12,496	−50%	0.555	74	174,913	151,863	−13%	0.733	0.656
Cotinine	9	24,064	23,610	−2%	0.976	73	115,926	89,836	−23%	0.260	0.729
Hydroxycotinine	9	1,089	214	−80%	0.240	73	5,465	3,011	−45%	0.221	0.482
All Areca alkaloids1	9	15,246	7,004	−54%	0.525	74	91,386	80,554	−12%	0.768	0.615
All tobacco alkaloids2	9	71,115	46,175	−35%	0.595	73	417,587	374,565	−10%	0.703	0.707

¹arecaidine + guvacine + guvacoline + arecoline

²nicotine + cotinine + hydroxycotinine

³geometric means (pg/mL)

⁴paired t-test n= number of participants

When data were separated by treatment arms at D22 (Figure 3) to test intervention effects all measured analyte levels decreased from BL in both the intervention (11–49%; n=74–75) and control (7–59%, n=81–83) groups. In the intervention group, the decreases reached significance only for nicotine (49%; p=0.007) and when all three measured tobacco alkaloids were combined (40%; p=0.002). In the control group, significant decreases were found for arecaidine (59%; p=0.021) and hydroxycotinine (51%; p=0.003). Detailed arecoline data in Supplementary Figures 1–6.

Figure 3.

Absolute numbers are geometric means. % numbers are relative changes to BL. None of the % changes to BL were significantly different between the intervention and control groups.

*p≤0.05; **p≤0.01; ***p≤0.001

When we used a proposed cutpoint of 30 pg/mL to distinguish D22 self-reported chewing status for all Areca alkaloids, the percentage classified as quitters at D22 was 48.0% for the intervention arm and 34.9% for the control arm (p=0.0968).

Discussion

Self-reports are the most commonly used approach to determine AN exposure and assess quitter prevalence in AN-based studies [14–18] due to their quick, straightforward, and inexpensive assessment methodology. However, self-reports can sometimes be unreliable [19]. Therefore, we sought to evaluate whether salivary alkaloids could be used to reliably bioverify self-reported BQ consumption.

Our most relevant finding was that combined salivary Areca alkaloid levels in self-reported quitters decreased relative to BL by 32%, but in self-reported chewers by only 9% (Figure 2), although the difference between the two groups was not significant (data not shown). This result was confirmed after data were stratified into treatment arms: quitters in the intervention arm showed a reduction of 26% while chewers reduced by only 2% (Table 1a); similarly, quitters in the control arm showed a reduction of 54% while chewers reduced by only 12% (Table 1b). This demonstrates the agreement between the self-reported chewer status and salivary bioverification with the four major salivary Areca alkaloids. This pattern was also seen for tobacco alkaloids, which suggest that these could possibly be helpful for bioverification for Class 2 chewers. Participants in the control group were also made aware of the health burdens and cancer risks of chewing via the information in the study pamphlet they received, which may have led to reduction in chewing in this group (“bogus pipeline effect”).

In a separate study examining the BENIT’s intervention effects exclusively from self-reports, AN/BQ chewing was significantly lower at D22 in the intervention group compared to the control group (39% vs 9%, p<0.01) However, in the current study using participant saliva alkaloid data, statistically significant differences between treatment arms from BL to D22 were absent (Figure 2), although differences were found between quitters and chewers, as noted above. The decreases in the alkaloids at D22 were stronger in the intervention than the control group and the percentage classified as quitters based on the proposed cutpoint at D22 was greater by 13 percentage points in the intervention than the control group. The reasons for this discrepancy could have been due to control participants only reducing (and not totally quitting) chewing—a finding we will evaluate quantitatively in a future and separate analysis. More likely, though, the disagreement is due to the limitations of the applied biomarkers. Foremost, Areca alkaloids disappear from saliva 10–12 hours after chewing [21]. For this reason, our study lacks the sensitivity needed to determine long-term chewing abstinence, and makes our measurements useful only heuristically. Given the limited detection window of ≤12 hours for salivary Areca alkaloids, it is possible that the saliva measurement did not accurately reflect AN use. For example, participants who did not quit (and, therefore, reported themselves as chewers) but temporarily stopped for the 12 hours preceding saliva collection, may have been misclassified as quitters based on salivary Areca alkaloid results (ie, ‘false negatives’). In contrast, and less likely, participants who did quit but had a short chewing episode in the 12 hours preceding saliva collection may have been misclassified as chewers based on salivary Areca alkaloid results (ie, ‘false positives’). This is illustrated in the percentage considered chewers at baseline of 40% based on the proposed D22 cutpoint for all Areca alkaloids combined, while all self-reported as regular users. To overcome these misclassifications, a future study with a much larger number of participants or with matrices that reflect longer-term AN exposure would be needed. In a separate study, we were able to improve the knowledge in Areca alkaloid pharmacokinetics when we explored the use of buccal cells and scalp hair as reliable matrices for long-term AN exposure (weeks in buccal cells to months in hair) using direct analysis in real time mass spectrometry [24, 25]. Unfortunately, the knowledge and technology to perform buccal cell or scalp hair analyses was not available to us during the BENIT.

Another study limitation is the lack of knowledge on the AN/BQ exposure beyond 12 hours, the time after which Areca alkaloids disappear from saliva [21]. This limitation casts doubt on the ability of our salivary biomarkers to declare someone a quitter versus a “reducer”, ie, someone who reduces chewing, but does not quit. To this end, the salivary bioverification of self-reported questionnaires highlights a difference between a dichotomous measure of quitting and a quantitative measure of chewing, such as Areca alkaloids levels and chewing frequency. Nonetheless, when quitters and chewers were compared with or without treatment stratification, our data shows encouraging results by demonstrating the usefulness of the applied biomarkers as bioverification tool. This will be an important consideration to implement in future and larger AN cessation efforts.

The measured tobacco related alkaloids in our study showed the same trend as the Areca related alkaloids, which confirms an intervention effect as the reduction was more pronounced in the intervention versus control groups. However, it is unknown whether the changes in the tobacco related alkaloid levels were due to changes in intake of tobacco in the participants’ BQs or other tobacco sources such as cigarette smoking, chewing tobacco, vaping, or other tobacco habits. Therefore, to accurately discuss tobacco related alkaloids in relation to BQ chewing cessation and their use as BQ biomarkers, we would need more comprehensive tobacco related information [26]. However, despite the unknown tobacco sources, knowledge about the changes in tobacco related alkaloids in the context of BQ chewing reduction is very helpful because of the addictive and carcinogenic properties of tobacco [27]. It is well known that tobacco use increases the risk of cancer and several oral potentially malignant disorders. This topic and its mechanisms has been covered in grave detail elsewhere [28, 29].

Data from some western Pacific countries where chewing is prevalent show survival at 5 years among oral cancer cases as low as 20%—a stark contrast to the average worldwide 5-year survival of over 50% [30]. In Guam, the incidence of mouth cancer in Micronesian individuals, a group who regularly chews AN, is almost three times higher than in Caucasians, a group who rarely consumes AN [31]. In addition, the duration and frequency of BQ use has been reported to increase the risk of developing oral submucous fibrosis, an irreversible condition with a high propensity for malignant transformation [7–9, 32]. This is of particular concern for individuals who begin chewing at a young age—as commonly occurs [3, 33–35]—due to the cumulative years of exposure and the longer life span. Thus, studies like the BENIT are of profound importance for this increasing and potentially deadly habit. The presented bioverification results of the BENIT will no doubt help to make considerable contributions towards improving cessation programs worldwide in hopes of decreasing the preventable health burden connected with AN/BQ chewing.

Conclusion

This is the first study to examine the use of objective biomarkers as a means to bioverify self-reported AN/BQ use in a cessation trial aimed at helping chewers quit. Salivary Areca and tobacco alkaloid concentrations compared well with self-reported chewing status with quitters having lower concentrations at D22 relative to BL. The presented data are encouraging for implementing this bioverification approach as a tool for short-term AN/BQ exposure in future AN cessation research.

Clinical Significance.

Results from our study can help in objectively evaluating compliance and program effectiveness for AN-based cessation programs worldwide.

Data availability statement

The data that support the findings of this study are available from the corresponding author, AAF, upon reasonable request.