Opium Use and the Risk of Liver Cancer: A Case–Control Study

In 2017, 35 million people worldwide suffered from drug use disorders (1). Opiate use (opium and heroin) is a significant health challenge in many countries because of adverse health consequences related to opiates abuse. World drug report has estimated that 29.2 million opiates users ages 15–64 (1). The early onset of opiates abuses leads to long-term exposure to the drugs, which raised a hypothesis of the carcinogenicity of opiates.

The most recent monograph working group on opium consumption has achieved strong consensus that opium use is a Group 1 human carcinogen for lung, larynx, and urinary bladder cancers by sufficient evidence (2). Many studies have associated opium use and various types of cancers such as esophagus, pancreas, stomach, and pharynx by epidemiologic studies, including cohorts and cases–controls, some other cancers like liver cancer are ignored (3–14). Limited experimental studies have shown that opium use could increase opium receptors. Increased receptor activity could lead to DNA damage and liver cancer (15).

Primary liver cancer (PLC), which afflicts approximately 841,000 new patients and causes 782,000 deaths annually, was predicted to have been the sixth most commonly diagnosed cancer and the fourth leading cause of cancer-related death worldwide in 2018 (16). Approximately 80% of liver cancer occurs in less developed countries (17, 18). Iran has a low incidence of liver cancer, and the age-adjusted incidence rate (ASR) in 2016 was 0.19 per 100,000 (19). Some regions of Iran showed higher ASRs than others, almost three times the difference between regions; meanwhile, the prevalence of common risk factors like hepatitis and alcohol drinking is almost consistent in the country. In a primary look at ecologic evidence, high ASR of liver cancer has occurred in the regions with a high prevalence of opium use (20, 21).

Alcohol trading and drinking are forbidden in Iran (22, 23); hence the prevalence of alcohol drinking (10%) is less than in other countries (24). The prevalence of opium use in Iran is the highest worldwide, the southeastern of Iran in particular (up to 20%) because Iran is the main trading pathways of opium from Middle-east to Europe-Golden Crescent (25, 26). By the ecologic concepts, it has been suggested to investigate opium use as a risk factor for liver cancer comprehensively (27).

Evaluating the possibility of the association needs well-designed epidemiological studies. To our knowledge, a cohort study in Golestan, Northern Iran, has reported a positive association between opium use and the risk of liver cancer, whereas a dose–response relationship was not evaluated. Furthermore, as the main confounder, cigarette smoking did not adjust, and the interaction of opium and cigarette smoking was not measured (28). The current study is a neighborhood case–control study in a highly prevalent area of opium use, which varies between 11% and 15% in the Southeast of Iran (29, 30), designed to evaluate the association between opium use and the risk of PLC.

A population-based case–control study was designed and carried out in a highly prevalent area of opium use worldwide-Southeastern of Iran.

Written informed consent was obtained from all subjects involved in the study and the study was conducted in accordance with the Declaration of Helsinki, and approved by ethical committee of Kerman University of Medical Sciences. The Ethic approval Code is IR.KMU.REC.1397.432.

A total of 117 pathologically confirmed patients with PLC were included in the study. The source of case selection was the comprehensive pathology-based cancer registry of Kerman University of Medical Sciences (KUMS) from January 2016 to December 2018. All PLCs were assumed as potential cases for our study. We excluded any history of hepatitis as a major risk factor for PLC. The lag time between cancer diagnoses and the interview had to be less than six months. Being Iranian and having Kerman residency for at least five years previous enrollment was another inclusion criteria

Having had IRB code, contact information, and cancer type was extracted from patients' medical records. After obtaining the consent via phone calls, trained interviewers set the dates and times for in-person interviews.

The neighbors were used as the source of controls. Two appropriate neighborhood controls were recruited. The pattern of control selection was described in detail elsewhere (8). Briefly, on the basis of the first two houses located on the right side of the case's house, two neighbors that met the inclusion criteria and consented to participate were enrolled and interviewed. The controls had to be matched by age (±5 years) and gender. The controls had not had a history of any other cancer diagnosis or hepatitis. Moreover, to exclude positive cases of hepatitis B and C, in addition to self-reporting by the participants, medical documents, pathology, and cancer registry reports were also examined. We ascertained that the controls did not suffer from a history of cancer by searching the control information on the cancer registry system. Other inclusion criteria were similar to cases except for PLC condition. In case of rejections or unavailability of control, the next neighbor was assessed for potential controls.

A standard and validated questionnaire was used (31). A family representative helped us fill out the questionnaire in case of any problems. To reduce information bias (recall bias) of opium use, a trained interviewer assured the participants that all information gathered anonymously and data would be stored confidential.

Detailed sections of the questionnaire were illustrated elsewhere (8). Our primary exposure of interestopium use "ever opium user" defined as those who had experienced opium use ever. Other opium metrics included age of initiation, duration of use (years), route of use (smoking and ingestion), type of opium, and the daily amount consumed (gram). Opium for daily consumption was estimated using the local unit "Nokhod" equal to 0.2 grams (32). The cumulative use of opium (gram-year) was measured by multiplying the intensity of use (gram) and duration of use (year). The dietary section of the questionnaire included red meats, white meats (chicken and fish), fruits and vegetables, and oils like frying oils. All the foodstuffs were categorized into low and high by scoring the frequency of use.

One of the main concerns about the link between opium and PLC is the reverse causality: Patients may consume opium to release pain (33). Therefore, for preventing the reverse causality, 17 participants who only had opium and drug use from the past until 2 years before the diagnosis of PLC or the enrollment into the study (for the control group), were excluded.

All analyses were done with STATA (version 14.0, Stata Corp.).

The datasets generated during and/or analyzed during the current study are not publicly available due to patient confidentiality but are available from the corresponding author on reasonable request.

A total of 117 patients with PLC and 234 neighborhood controls were analyzed. Most participants were men (68.09%), and subjects mostly were married (96.01%). All participants were over 55 years old. Controls and cases' mean age was 62.70 ± 13.22 and 64.29 ± 13.43 (P = 0.85). Patients were more cigarette smokers (38.46% vs. 20.94%) and opium users (18.10% vs. 9.83) than controls. About 2 (1.71%) of the cases and 3 (1.28%) of the controls had a history of alcohol drinking (Table 1).

Table 2 showed the association of opium use and the risk of primary liver cancer. The risk of PLC was six times higher among opium users than controls (AOR, 6.50; 95% CI, 2.87–13.44). Those who consumed more than 1 gr/d were at risk of PLC (AOR, 10.70; 95% CI, 3.99–28.72).

Duration of opium use was significantly associated with the risk of PLC. Subjects who used opium for more than 21 years were about 11 times at higher risk of PLC (AOR, 11.65; 95% CI, 4.43–30.67).

The cumulative use of opium, as we expected it, was increasing to about 11 times among higher levels of cumulative use of opium, >30 gr-year (AOR, 11.00; 95% CI, 3.83–31.58).

Early start of opium use, less than 50 years old, increased the risk of PLC to AOR, 11.31; 95% CI, 4.41–29.0.

The opium smoking method increased the odds of PLC more than 5 times (AOR, 5.94; 95% CI, 2.47–14. 25) Oral ingestion leads to an extremely high risk of PLC (AOR, 30.57; 95% CI, 3.45–270.37).

Raw opium use was prevalent in both groups. None of the controls used refined opium.

The risk of PLC increased to 4.67 (95% CI, 2. 07–10.0) in raw opium users than never opium users. Cigarette smoking in the adjusted model was not associated with the risk of PLC.

The results showed that on an additive scale, 79% of the association goes through the path of opium consumption (ever/never) and the other 21% of the association goes through the path of other risk factors. In addition, on a multiplicative scale, 90% of the relationship goes through the path of opium use (ever/never) and 10% through the path of other behavioral risk factors.

Moreover, the results showed that the consumption of opium ≤1gr/d on an additive scale causes 75% of the odds of getting PLC, and on a multiplicative scale 85% of the odds of getting PLC. Although consumption of >1gr/d on an additive scale causes 79% of the odds of getting cancer and on a multiplicative scale 92% of the odds of getting PLC.

On an additive scale, the duration of opium use of ≤21 years predicts 68% of the odds of getting PLC and 22% of the odds of getting PLCs predicted by other behavioral risk factors. Although the duration of use of opium ≤21 years on a multiplicative scale causes 77% of the odds of developing PLC, other behavioral risk factors predict only 23% of the OR. Also, on the additive scale, the duration of opium use >21 years caused 80% of the odds of getting PLC, and on the multiplicative scale, 93% of the odds of getting PLC.

Cumulative use of opium ≤30 gr-year caused 72% of the odds of getting PLC on an additive scale and 84% of the odds of getting PLC on a multiplicative scale. On an additive scale, the cumulative consumption of opium >30 gr-year caused 78% of the odds of getting PLC, and on a multiplicative scale, 91% of the odds of getting PLC.

Also, the results showed that the age of starting opium consumption ≤50 years predicts 79% of the OR on the additive scale and 92% of the chance ratio on the multiplicative scale.

In addition, the age of starting opium use <50 years predicts 45% of the OR in the additive scale and 48% of the OR in the multiplicative scale.

Also, the results showed that, on an additive scale, the smoking method of opium use predicted 73% of the odds of getting PLC, and on a multiplicative scale, it predicted 87% of the odds of getting PLC. In addition, the ingestion method of opium consumption includes 94% on additive scale and 98% on multiplicative scale of the odds of PLC.

Furthermore, on an additive scale, 69% of association goes through the path of raw opium and 31% through the path of other behavioral risk factors. Also, the raw type of opium in the multiplicative scale includes 84% of the association (Table 2).

Furthermore, in the present study, the interactions between cigarette smoking (ever use) and opium consumption (ever use or cumulative dose) were not statistically significant (P = 0.3 and 0.7, respectively).

To evaluate the stability of the estimate of cumulative consumption OR, we perform a sensitivity analysis of the association between cumulative opium use and PLC, and we observed that the OR changed by a maximum of 5% (Supplementary Table S1). This indicates stability and robustness in the estimated OR. Moreover, sensitivity analysis of self-reporting of opium use was significant after reducing the number of opium users to 50% of cases and controls. Moreover performing a sensitivity analysis between the cumulative use of opium and the occurrence of PLC showed that the estimated OR changes by a maximum of 5%, and this indicates the stability and robustness of the estimate.

In this study, we found that opium use was highly associated with the risk of PLC. We found a strong increase in the risk of PLC associated with the early onset of opium use and oral ingestion of opium. Moreover, the dose–response relationship between opium and PLC was convincingly demonstrated. Our results were in line with one of the largest cohort studies in Iran, the Golestan cohort study; the incidence rate of PLC was about 2.5 times higher among opium users than non-users (AOR, 2.46; 95% CI, 1.23–4.95; ref. 28). PLC mortality increased to six times among opioid-dependent participants registered in the opioid substitution therapy (OST) program (34).

Our data would suggest that confounding behavioral factors reduce the PLC ORs approximately 21% on additive scale and from 10% on multiplicative scale of the predicted effect of opium use on PLC, meaning that other behavioral factors have limited potential for reducing opium use in PLC ORs, and it is noteworthy to consider opium use as a substantial element of PLC disease in our society if there are no residual confounding factors.

On the basis of the recent publication in nature genetic it was shown that mutational profiles were similar across all countries. In addition, association between specific mutational signature and esophageal squamous cell carcinoma were defined for some risk factors like tobacco, alcohol, and opium. In this study, opium consumption was identified as a mutational signature associated with ESCC. However, the researchers were unable to fully distinguish it from other signatures with overlapping contexts. Further in vitro studies are recommended to determine the signature effect of opium as a mutagen on cancers (35).

Aside from the biological plausibility of opium and the etiology of PLC, a brief look at the incidence rate of PLCin Iran and comparing it with the prevalence of opium use could further clarify the association. The current study was carried out in Kerman, Southeastern Iran, where the prevalence of opium use has been estimated to be about 10% in 15 to 75 years the population (29); in addition, the prevalence of risk factors such as hepatitis B and alcohol drinking was similar to other parts of Iran (36). Therefore, another risk factor like opium might be considered. ASR of PLC of Kerman was 0.7, whereas the liver ASR of Iran was less than 0.2 (19, 37). Three folds higher ASR of the high prevalence area of opium use suggests that opium could play an important role in PLC etiology.

Opium use is a stigma behavior in Iran. The major challenge of our study was to collect reliable data and avoid misclassification bias. Previous studies have shown that neighbors could be a suitable source for control selection for studies with sensitive questions like drug abuse; nonetheless, neighbors suffered from a sort of under-reporting (38, 39). We trained the interviewer to deal with the stigma situation by explaining the confidentiality of contact information, data gathering, and storage. Patients are less under-reported for drug abuse than controls because doctors have to know about drug abuse before starting treatment to reduce withdrawal syndrome. Moreover, the association was valid in sensitivity analysis of opium self-reporting, even by half positive opium reporting in both cases and controls.

OR could be an appropriate measure of the risk ratio in rare diseases, including cancer. So, the excess odds explained in the current study would be a proper estimate of excess risk explain.

We chose neighborhood controls for this study because they were a better choice than hospital controls because opium use causes a variety of known and unknown diseases (38). Furthermore, friend/family controls had the disadvantage of being overmatched for opium use, as these habits predominate in specific social networks.

This study has other limitations. Although we tried to adjust major risk factors, some were missed, like the history of hepatitis B or C, cirrhosis, family history of cancer, and exposure to Aflatoxin. Drug abusers are more likely to have criminal records and a higher prevalence of syringe sharing and hepatitis incidence (40). Broad hepatitis vaccination has been done since the 1990s; nevertheless, future studies must seriously consider the infections. However, on the basis of the systematic and met regression analysis that was done in Iran from 2000 to 2016 it was shown that the age standardized hepatitis B prevalence in Iran decreased from 3.92% in 2000 to 1.09% in 2016. In addition, the annual prevalence of hepatitis in Kerman was 3.22, and 4.44 in 2000, and this prevalence declined to 0.88 and 0.81 in 2016. Also, Iran has shown a low prevalence of hepatitis B compared with its neighboring countries. Therefore, we can conclude that the very low prevalence of hepatitis B in Kerman especially in recent years can decline the magnitude effect of hepatitis for inducing PLC (41).

Long-term alcohol drinking has been linked to an increased risk of PLC, but the Islamic culture of Iran's alcohol drinking is deficient (24). Therefore, Iran is a unique place to study other risk factors (such as opioids) of PLC (41). The use of opioids like hydrocodone, oxycodone, codeine, tramadol, and fentanylize increasing for medical and non-medical use globally (42); accordingly, severe health consequences such as PLC have to be investigated widely, and a consortium may pave the way to study this health challenge comprehensively.

In conclusion, the results of this study indicate that opium use dramatically increased the risk of PLC.

No disclosures were reported by all authors.

M. Marzban: Conceptualization, resources, data curation, software, formal analysis, funding acquisition, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing. E. Mohebbi: Conceptualization, data curation, software, formal analysis, investigation, visualization, methodology, writing–original draft. A.A. Haghdoost: Investigation, methodology. M. Aryaie: Visualization, writing–original draft, writing–review and editing. M.J. Zahedi: Investigation, methodology, writing–original draft. Z. Khazaei: Data curation, investigation, visualization, methodology, writing–original draft. M. Gholizade: Conceptualization, data curation, software, formal analysis, investigation, visualization, methodology, writing–original draft. A. Naghibzadeh-Tahami: Conceptualization, resources, data curation, software, formal analysis, supervision, funding acquisition, validation, investigation, visualization, methodology, writing–original draft, project administration, writing–review and editing.

This study was partly funded by Clinical Research Development Unit, Shafa Hospital of Kerman University of Medical Sciences (KUMS) through grant number 97000899 (A. Naghibzadeh-Tahami). The funding bodies were not involved in the study design, study implementation, or writing the article not-for-profit sectors. The authors would like to express their gratitude to all study participants for their cooperation in conducting face-to-face interviews. We also thank Pouya Dehghani, who interviewed the participants. The authors also would like to thank the Physiology Research Center at Kerman University of Medical Sciences (KUMS) for financially supporting this research through grant number 97000899.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.