The spectrum of opportunistic infections and malignancies among women on antiretroviral therapy in Ethiopia

ABSTRACT Background: Women are disproportionally affected by HIV and are at greater risk of OIs and malignancies. This study investigated the spectrum of OIs and malignancies among women living with HIV (WLHIV) in Ethiopia. Methods: A retrospective longitudinal study was conducted among 3,817 WLHIV on Antiretroviral Treatment (ART) from 2007 to 2019 in Ethiopia. Data on OIs, malignancies, and clinical characteristics were retrieved from medical records. Prevalence, incidence, and mortality rates were measured over 13 years of follow-up. The Kaplan-Meier survival curve and multivariable Cox-proportional hazard regression were used to estimate incidence and identify predictors of OIs and malignancies, respectively. Result: The pooled prevalence of OIs or malignancies was 47%. OIs accounted for 39% and malignancies were 23.3%. The incidence of OIs and malignancies were 19.4 (95%CI:9.2-29.5) and 11.5 (95%CI:2.2-20.8) per 100 person-years of observation, respectively. Incidence of candidiasis was 14.1 per 100 person-years followed by Lymphoma (11.9), Cervical Cancer (11.5), and TB (10.9). The risk of OIs was higher in patients with baseline CD4 count ≤500 cells/mm3 [AHR (95%CI): 2.0 (1.8-2.3)] and VL>1000 copies/ml [AHR (95%CI): 1.5 (1.1-1.9)]. Overall, 24.6% died, and of them, 62.2% had a history of OIs or malignancies. The risk of death was higher among WLHIV with a history of Toxoplasmosis, TB and PCP. Conclusion: OIs and malignancies was recently increasing and two-thirds of all deaths were associated with these co-morbidities. The high rates of OIs and malignancies were attributed to poor treatment outcomes. The finding calls to enhance the diagnosis and treatment of these co-morbidities. Abbreviations AIDS: acquired immune deficiency syndrome; CI: confidence interval; EPHI: Ethiopian Public Health Institute; HAART: highly active antiretroviral therapy; HIV: human immunodeficiency virus; HR: hazard ratio; Mg/dl: milligram per deciliter; TB: tuberculosis; PCP: pneumocystis carinii pneumonia; ZJU: Zhejiang University


Background
Opportunistic infections (OIs) and malignancies are common in individuals with weakened immune systems, mainly among people living with HIV (PLWHIV) and malnourishment [1][2][3].These conditions contribute to more than 90% of HIV/AIDSrelated deaths [1].Women were also reported to be disproportionally affected by OIs [2].Moreover, malignancies are the most common comorbidities among women in low-and middle-income countries (LMICs), particularly cervical cancer and lymphoma [3][4][5].Moreover, women are disproportionally affected by the HIV epidemic with a prevalence of 4.2% compared to males (1.2%) in urban Ethiopia [6].Malnutrition among women was also reported to be significantly higher (33.2%) than in males (19.3%) [7].
Despite advances in HIV diagnosis and treatment, OIs, mainly Tuberculosis (TB), oral candidiasis, Herpes zoster, Toxoplasmosis, and Pneumocystis Pneumonia (PCP), remain major causes of morbidity and mortality in HIV/AIDS patients in LMICs [8].Recent studies in Ethiopia have shown an increasing trend of OIs among PLHIV [9,10].On the other hand, malignancies have also been among the leading causes of mortality in PLHIV which accounted for 25%-40%.The commonest malignancies among WLHIV are Lymphoma and cervical cancer [11].Cervical cancer is the second leading cause of cancerrelated illness and deaths among women with an estimated 570,000 cases and 311,000 deaths per year [5], and about 90% of cases and deaths occur in LMICs [12].Ethiopia is among the top ten countries globally and second in Eastern Africa with over six thousand cases and four thousand deaths annually [9].HIV is associated with Human Papilloma Virus (HPV) infection augments the progression, transformation, and aggressiveness of premalignant lesions due to HPV infection [9].
Hence, early detection of OIs, cervical cancer, and lymphoma among PLHIV is essential to prevent complications, improve treatment outcomes, and ultimately prolong survival.In recent years, efforts have been made to integrate the diagnosis and treatment of OIs and malignancies in HIV care and treatment in Ethiopia.Moreover, the country has been working on screening cervical cancer using Visual Inspection using Acetic Acid (VIA) since 2016 and has been moving towards universal HPV vaccination for young adults (15-24 years old) since 2020 [10].However, there is no clear programme strategy for screening other non-communicable diseases among PLHIV in Ethiopia.The country also had a guideline on integrating TB-HIV services during the clinical management of PLHIV.This includes diagnosis of active TB for PLHIV and providing INH/IPT and CPT prophylactic.
Despite these efforts, there is limited evidence on the magnitude of OIs and malignancies among WLHIV in Ethiopia.Hence, this study aimed to describe the spectrum (i.e.incidence and prevalence) of OIs and malignancies and associated mortality among WLHIV in Ethiopia.

Study design and setting
This study was part of a nationwide cohort study on HIV-1 treatment and drug resistance at 63 health facilities in Ethiopia between 2007 and 2019 [8].Of the 63 facilities, 20 referral hospitals and university hospitals that had OIs and malignancy diagnosis services were included in this analysis.

Population, sample size, and sampling
We included all Adult WLHIV (≥15 years old) who were enrolled between January and December 2007 and on follow-up for any length of time until December 2019 at the 20 hospitals with diagnosis services for OIs and malignancy.We excluded WLHIV who were diagnosed with any OIs and malignancies at baseline screening.We also excluded patients with incomplete records of baseline information such as CD4 count, WHO clinical stage, date of Highly Active Antiretroviral Therapy (HAART) initiation, functional status, and status of OIs or malignancies.Loss to follow-up was also excluded from the analysis as of the date of their classification as lost.Moreover, accordingly, a total of 3,817 WLHIV were found eligible and included in the analysis (Figure 1).

Diagnosis of OIs and malignancies
At recruitment for the study, we evaluated and excluded those who were not eligible for this study and included as much as possible baseline information.We also have evidence of time to diagnosis.However, our analysis started from the time of HAART initiation since the test and treatment were not implemented by that time and the eligibility criterion for HAART initiation was a CD4 count of <500 cells/mm 3 .This may imply the high burden of OIs and malignancies at an early stage of HAART initiation.

Measurement
Incidence: Time from ART initiation to the occurrence of any OIs or malignancies among WLHIV from recruitment until death or lost-follow-up or end-of-follow-up.
Prevalence: the occurrence of any OIs or malignancies among WLHIV at any time during the follow-up period.
Mortality: death of women with a history of OIs or HIV-associated malignancies divided by overall death.
Lost to follow-up: Patients missing their follow-up visits according to their appointment schedule for more than 3 months as recorded on the appointment schedule.

Data collection
Data were extracted from medical and electronic records using an abstraction tool designed for this purpose and adapted from the national HIV care and ART follow-up intake records from March to May 2020.We used the Open Data Kit (ODK) programmed in tablet computers and integrated into the Ethiopian Public Health Institute (EPHI) server for real-time data capture.Data were extracted on occurrences of OIs, malignancies, deaths, and explanatory variables including socio-demographic (i.e.age, sex, residence, marital status, and occupation); Clinical and laboratory (i.e.clinical stage, CD4 count, and functional status), and medication-related (i.e.HAART regimen and adherence).Data completeness and consistency were double-hecked on 15% of randomly included records.The predictor variables had been recorded at baseline and occurrences of OIs, malignancies, and death were registered during follow-up period.

Data analysis
Descriptive statistics were employed to summarize the incidence and percentage of OIs and malignancies, and frequencies and median/mean to summarize key socio-demographic and clinical characteristics.The Kaplan-Meier curve was used to demonstrate the incidence of OIs and malignancies per 100 person-year observations.A cox-proportional hazard regression model was fitted to identify the predictors of OIs and malignancies among WLHIV.Variables with pvalues ≤0.20 in bivariate analysis were included in the final multivariable model.In addition, a generalized Log rank test was used to compare the OI or malignancy survival time between different categorical predictor variables.Adjusted Hazard Ratio (AHR) with a 95% confidence interval was used to determine effect sizes at p < 0.05.STATA Version 17 was used for analysis.

Socio-demographic and clinical characteristics
A total of 3817 WLHIV were included in the analysis.The mean age of participants at baseline was 38 (±10) years old.Majorities were urban by residence (93.2%), ever married (89.5%), and aged 26-45 years (74.5%).At baseline, 63.8% had a CD4 count of >500 cells/ mm 3 and were on WHO clinical stage I (91%) (Supplementary Table 1).

Trends of OIs and HIV-associated malignancies
The overall prevalence of OIs during the first year of follow-up was 62.8% which declined overtime until the nineth year of follow-up (19.1%) and then significantly increased (P < 0.001) to 39% by the last year of follow-up.The prevalence of oral candidiasis declined from 28.0% at baseline to 12.2% by 2019, PCP from 9.5% to 3.3%, and Toxoplasmosis from 2.9% to 2.2%.Herpes Zoster was relatively constant (i.e.21.7% by 2007 vs. 24.4% by 2019).TB increased from 20.1% in 2007 to 38.9% in 2019, and lymphoma increased from 27.6% to 33.3% (Table 2 and Figure 2).

Discussion
This study showed that the pooled prevalence of OIs or malignancies among WLHIV during the 13 years of follow-up was 47% (39% for OIs and 23.3% for malignancies).The most prevalent OIs were oral candidiasis (14.2%),TB (11.3%) and Herpes zoster (9.6%), whereas the prevalence of cervical cancer and Lymphoma was 11.9% and 13.3%, respectively.The Incidence of OIs or malignancies was 19.7 per 100 person-years observation, respectively.Oral candidiasis was the most incident case with 14 per 100 personyears followed by lymphoma (12), Cervical Cancer, and TB (11).Patients with CD4 count <500 cells/ mm 3 and VL > 1000 copies/ml had a higher risk of acquiring OIs and malignancies.A quarter of patients died during follow-up and significantly associated with Toxoplasmosis, TB, and PCP.
The overall prevalence of OIs in the current study was almost double the findings from Northwest Ethiopia [14] but much lower than another report from Northeast Ethiopia [15].Moreover, the increasing trends of oral candidiasis and Tuberculosis after ART initiation were in line with the report in Northeast Ethiopia [15].On the other hand, the incidence of OIs in this study was higher than previous reports in Northeast [15] and northcentral Ethiopia [16], China [17], and South Korea [18], comparable with a report in Senegal [8], and lower compared to the finding in India [19].This variation might be attributable to variations in length of follow-up, sample sizes, study population, and socio-economic and healthcare settings, for example, the previous studies on Ethiopia were conducted on single health facilities.Specifically, the prevalence of oral candidiasis (14.2%) and Herpes zoster (9.6%) in this study was relatively lower compared to those who were HAART naïve in LMIC which was reported to be 19.1% [95% CI, 13.0%-27.3%]and 14.4% [95% CI, 6.7%-13.2%][8], respectively.This could be explained by the importance of HAART in reducing OIs.Moreover, 11.3% prevalence of TB in this study was significantly higher compared to prevalence in the general population in Ethiopia which was reported to be 0.19% [95% CI: 0.12%-0.28%][20,21].This highlights a 6-fold risk of TB among PLHIV compared to the general population in the country.
The overall prevalence of malignancies was relatively higher compared to the findings [4,8,11,18,20,21].A relatively higher prevalence of cervical cancer in this study could be explained by the limited availability of HPV vaccination, cervical cancer screening programmes, and a high prevalence of human papillomavirus (HPV) infection among HIVinfected women in Ethiopia [6,10].A systematic review estimated a higher prevalence of cervical cancer and reported the risk of cervical cancer increased in WLHIV (RR = 6.07, 95% CI = 4.40,8.37).Globally, 5.8% (95% CI = 4.6,7.3) of new cervical cancer cases in 2018 (33,000 new cases, 95% CI = 26,0,42,0) were diagnosed in and 4.9% (95% CI = 3.6,6.4)were attributable to HIV infection (28,000 new cases, 20,000-36,000).Moreover, the PEPFAR Ethiopia reported     age-standardized incidence rate of HPV among the general population in Ethiopia was 21.5 per 100,000 person-years equivalent to 9.6% prevalence by 2020 [22].Our report is significantly higher compared to these estimates which could be due to the fact that our study population had a 13 years of follow-up.
The previous report also highlighted most affected regions by cervical cancer were southern Africa and Eastern Africa.In southern Africa, 63.8% of WLHIV had cervical cancer, compared to 27.4% in Eastern Africa [9].This was also relatively higher compared to our report which could be explained by the scope of the study which covered East African countries and may also be due to the fact that cervical cancer screening is yet a challenge and under-estimated by the different studies.Our finding, therefore, calls for strengthening cervical cancer screening among WLHIV, particularly among those who had more than 10 years of HAART experience women who had the highest incidence rate.The overall prevalence of Lymphoma among WLHIV in Ethiopia was 13.26%.Moreover, Lymphoma among the general population in Ethiopia accounted for 4.9% [23].This report was also higher than a systematic review conducted in India (6.5%) [24].In this study, Lymphoma among elderly WLHIV was high which could be explained by the immune deterioration of the elderly that could exacerbate the development of lymphoma and other malignancies.This finding highlights the importance of lymphoma diagnosis among the geriatric population.
Our analysis shows that WLHIV with CD4 count <500/mm 3 had a two-fold higher risk of developing OIs which is consistent with previous studies [8][9][10]17,18,25,26].Similarly, a higher risk of OIs among those with VL > 1000 copies/ml [17,25,27].The result of the association between CD4 count and incident OIs confirms the known biological relationship between low CD4 count and OIs [28].The higher risks of malignancies among patients with CD4 count <500/mm 3 and VL > 1000 copies/ml were also supported by many studies [4,11,18,24,29].Hence, the impact of OIs and malignancies on treatment outcomes is inevitable.The re-increase of OIs and malignancies as of 2014 could be explained by a war in the country which limited access to health services and follow-up of PLHIV to the health facilities.This limited not only the HIV care and treatment but also factors in late diagnosis.The other reason may be, the limited attention on these co-morbidities in recent years.The rate of mortality among WLHIV was consistent with previous studies in Ethiopia [30,31], and significantly higher rates of death in patients with multiple infections were also supported by various studies [11,25,32,33].However, this study reported a two-third mortality associated with OIs or malignancies which was significantly high.This highlights the significant impact of these comorbidities among WLHIV in Ethiopia.The higher risks of death among WLHIV diagnosed with Toxoplasmosis and TB were consistent with previous studies [17,[34][35][36][37].
Due to the limited information in the recordings, the different stages of cervical cancer and the types of lymphoma were not characterized.Nearly 20% of records were excluded due to lost-to-follow-up or incomplete records.This would affect the generalizability of findings since the missed cases would have special features that would affect the occurrence of OIs and malignancies.Moreover, this study was conducted at referral hospitals and university hospitals where most patients are urban and relatively high economic compared to rural WLHIV.The facilities had limited capacity during the earlier times to diagnose OIs and HIV-related malignancies and hence misclassification or misdiagnosis might happen.Since this study was concluded by the end of 2019, the burden of OIs and malignancies during after the era of COVID-19 was not part of this study.

Conclusion
Nearly half of WLHIV develop either OIs or malignancies while the incidence rate was also increasing, particularly for TB and Toxoplasmosis.A quarter of WLHIV died during the 13 years of follow-up, of whom two-thirdss were associated with OIs or malignancies.The high rates of mortality associated with opportunistic infections were attributed to immunologic and virologic failures.The finding calls to enhance the diagnosis and treatment of OIs and malignancies particularly TB, Toxoplasmosis, and cervical cancer during the clinical management of WLHIV.

Operational definition
OIs: common opportunistic infections (i.e.TB, Candidiasis, Toxoplasmosis, PCP, and Herpes zoster).TB was diagnosed using AFB microscopic or GeneXpert.Moreover, oral candidiasis, Toxoplasmosis, PCP, and Herpes zoster were defined per the clinical definitions according to the national comprehensive ART guideline on Ethiopia [13].
Malignancies: common malignancies among WLHIV (Cervical cancer and Lymphoma).Cervical cancer was determined using Visual Inspection with Acetic Acid (VIA) or PAP smear complemented with clinical evidence as recorded by the attending physician.Lymphoma was also defined per the clinical definitions according to the national comprehensive ART guideline on Ethiopia [13].
Lost to follow-up: Patients missing their follow-up visits according to their appointment schedule for more than 3 months as recorded on the appointment schedule.
WHO clinical stages, CD4 count, and VL suppression were described as per the national compressive ART guideline [13].

Figure 2 .
Figure 2. Trends of Incidence and prevalence of Opportunistic infection and HIV-associated malignancies among WLHIV in Ethiopia (2007-2019): (A) Incidence of OIs and HIV-associated malignancies (per 100-person year observation) (B) Trends of prevalence OIs and HIV-associated malignancies.

Figure 3 .
Figure 3. Mortality associated with OIs and malignancies among WLHIV in Ethiopia (2007-2019).(A) Incidence of OIs and HIVassociated malignancies (B) Proportion of mortality disaggregated by OIs and malignancies (C) Trends of Mortality associated with OIs and malignancies.

Figure 4 .
Figure 4. OIs or malignancies associated with mortality among WLHIV in Ethiopia by their importance.

Table 1 .
Prevalence of OIs and HIV-associated malignancies disaggregated by clinical characteristics among WLHIV in Ethiopia (2007-2019).

Table 4 .
The risk of OIs and HIV-associated malignancies on mortality among WLHIV in Ethiopia (2007-2019).