Proles of 2-Drug Regimen in People Living with HIV-1 in A Real World Setting: A Large-Scale Medical Claim Database Analysis in Japan

Background Regimen simplication to 2-drug antiretroviral therapy (2-ART) may address potential tolerability issues, increase adherence, and reduce toxicity and potential drug-drug-interactions among people living with HIV-1 (PLWH). However, real-world treatment patterns and patient proles associated with 2-ART are unclear. Methods This retrospective observational cohort study employed a large-scale medical claim database of Japanese hospitals to extract data on 4,293 PLWH aged ≥ 18 years with diagnosis of HIV and treated with any ART regimens between April 2008 and April 2019. A 2-ART cohort was compared with a 3-drug antiretroviral therapy (3-ART) cohort in terms of patient characteristics, comorbid conditions, and treatment patterns. Treatment switching rates were calculated for each cohort followed by sensitivity analysis to conrm the robustness of the ndings. Results There were 94 patients identied in the 2-ART cohort. Compared to the standard 3-ART cohort (n=3,993), the 2-ART cohort was older (mean age 54.4 vs 43.4 years), with a lower proportion of males (87.2% vs 93.8%), higher Charlson Comorbidity Index (CCI) (mean score 6.9 vs 5.3), more co-medications (mean 8.3 vs 5.0), and a higher percentage of AIDS-dening conditions (66.0% vs 42.8%). The most common 2-ART were protease inhibitor (PI) + integrase strand transfer inhibitor (INSTI) and non-nucleoside reverse transcriptase inhibitor (NNRTI) + INSTI (33.0% and 31.9%, respectively). Overall, most of the regimens were nucleoside reverse transcriptase inhibitor (NRTI)-sparing (71.3%), with a decreasing trend over time (76.2% to 70.2%). ART regimen switch occurred more often in the 2-ART cohort than in the 3-ART cohort (33.0% vs 21.2%).


Introduction
Antiretroviral therapy (ART) regimens have been credited with increasing the survival of people living with HIV-1 (PLWH) [1]. Three-drug ART regimens (3-ART), which consist of a backbone with two nucleoside reverse transcriptase inhibitors (NRTIs) plus an anchor drug with or without a booster, have become the standard of care (SoC) and treatment for HIV-1 infection [2,3]. While advances in ART have been credited with improving survival by decades [4,5], recent evidence of long-term adverse events (AEs) with 3-ART have been documented [6], particularly in terms of renal and bone toxicity [7][8][9][10]. Such poor outcomes have prompted development of drug-sparing strategies with a simpli ed 2-drug ART regimen (2-ART) in attempt to reduce its negative impact.
In contrast, multi-center clinical trials of 2-ART (dolutegravir plus lamivudine and dolutegravir plus rilpivirine) have demonstrated to have non-inferior e cacy as well as safety compared to 3-ART [11,12].
Based on the positive impact, there is an increasing diversity of therapies with a 2-drug combination of integrase strand transfer inhibitors (INSTI) plus protease inhibitors (PI) or non-nucleoside reverse transcriptase inhibitors (NNRTI). More ART regimens are being developed and used in real-world settings to tailor to the long-term complexities related to a patient's genetic background, treatment history, and comorbidities [13].
In Japan, ART treatment uptake and clinical success among diagnosed HIV patients, in terms of viral suppression, has been shown to be excellent and in line with UNAIDS/WHO targets of 90% [3,14,15].
However, Japanese PLWH have been demonstrated to have unique comorbidities and co-medication patterns [16]. There is limited evidence that switching to 2-ART may improve outcomes [17]. While the practice of regimen switching and reduction is an increasing global trend, the pro le and number of patients receiving 2-ART in real-world practice are not well understood in Japan. This study aimed to characterize patients who received 2-ART and provide an overview of ART treatment patterns in Japan.

Study design and data source
This study was a retrospective cohort analysis of ART treatment patterns and patient characteristics among adult patients living with HIV on 2-ART compared to 3-ART in Japan. Data were extracted from a hospital-based medical claims database maintained by Medical Data Vision Co., Ltd. (MDV, Tokyo, Japan). The data is from all hospitals included in the MDV hospital panel, which covers 374 acute phase hospitals across regions (approximately 22% of acute phase hospitals in Japan) and roughly 25 million people. The MDV database includes treatment information, including antiretrovirals used, co-medications, and comorbidities, etc., but no laboratory results such as CD4 counts or plasma viral load.

Study Population
PLWH were included in the analyses if they had a diagnosis of HIV (ICD-10: B20-B24), had been treated with 2-or 3-ART, and were aged ≥ 18 years. The patient identi cation period for treatment index dates was from the rst date available in the database (April 2008) to the last available (April 2019). A patient was considered to be on a 2-ART if they were prescribed any two antiretrovirals on the same day, and the rst date of their 2-ART was the index date for this cohort. Patients who did not have any 2-ART during their patient record were eligible for the 3-ART cohort if they were prescribed three antiretrovirals on the same day. The rst date of their 3-ART was considered to be the index date for this cohort. Patients were excluded if they were treated with a C-C chemokine receptor type 5 (CCR5) antagonist or two antiretrovirals in the same class as their index 2-ART because they are not SoC. Additionally, patients treated with any regimens outside of 1) INSTI + 2NRTI, 2) PI + 2NRTI, or 3) NNRTI + 2NRTI during their index 3-ART were excluded as they are also not SoC.
Demographic and clinical characteristics for the 2-ART cohort were strati ed by three time periods to assess possible temporal changes due to novel drug access in Japan: April 1, 2008 to September 30, 2014 (representative of 1st generation INSTI availability, mainly raltegravir), October 1, 2014 to June 30, 2016 (representative of 2nd generation INSTI availability, including dolutegravir), and July 1, 2016 to April 30, 2019 (last available data period, representative of tenofovir alafenamide (TAF) availability as a novel prodrug of tenofovir to reduce toxicities). All drug classes and combinations were also tabulated for all patients. The average gap (days) between 2-ART and the total average duration of treatment were calculated from regimen initiation to last ART prescription in the database. Time to index regimen for the 2-ART cohort was calculated as the time from patients' rst ART (prior to 2-drug initiation) to their index date.

Patient pro les
The analyses compared patient characteristics on 2-ART with those with 3-ART. Patient pro les consisted of demographics (age at index, gender, year of index date), co-medications, and comorbidities (hemophilia, AIDS-de ning conditions, modi ed Charlson Comorbidity Index (CCI)). Modi ed CCI is severity indicator which is able to predict one-year mortality among hospitalized patients by weighting 17 different comorbid conditions (e.g. myocardial infarction, congestive heart failure), and has been validated using Japanese data [18]. Additionally, all 2-and 3-drug combinations observed for each cohort were tabulated. These combinations were described by drug class: INSTI, PI, NNRTI, and NRTI. Regimens that did not contain any NRTIs were considered to be NRTI-sparing regimens in alignment with the de nition of U.S. Department of Health and Human Services [19] .
ART regimen switch Two-and 3-ART switch patterns were also assessed. Patients were considered to have switched their drug regimen if they had any change to their index regimen, such as an addition, removal, or exchange of a drug. For patients on 3-ART, they cannot switch to a 2-ART at any point by de nition, since any evidence of 2-ART in a patient's record would classify that patient in the 2-ART cohort. In order to capture true switches, regimen changes were only counted as switches if they lasted for more than 90 days. Both patients on 2-and 3-ART were considered to have switched even if they remained on the same number of drugs, as long as a change in any component of their regimen as present. Speci c transitions were not considered switches: the same regimen with only tenofovir disoproxil fumarate (TDF) switched to TAF, the same regimen with changes in dosing, or the same regimen with an added/switched booster. Switch rate was de ned as number of patients who switched divided by total person years. Total person-years for treatment-switches was time from index treatment initiation to the rst treatment switch, whereas for patients who did not switch it was de ned as time from the rst treatment initiation to the last ART prescription during the study period. The switch rate and time-to-next-switch were calculated from the index date to the patient's next treatment switch, the last ART claim in the database or end of the study period. The switch rate observed in among both cohorts were compared.
Finally, a sensitivity analysis of switching was performed to ensure the robustness of treatment pattern ndings. As described in Fig. 1, only patients on 2-ART with at least one antiretroviral drug prior to their index date and patients on 3-ART with at least 1 switch after their index date were included to ensure the assessment of treatment experienced patients. In addition, only patients with at least one year of followup data were included to ensure that potential switches could be captured in the data. In the switch sensitivity analyses, the switch index date was set to patient's rst switch for the 3-ART cohort and the switch rate was calculated from the time of the rst switch to their next switch.

Statistical Analysis
All statistical analysis was completed on SAS version 9.4 (SAS Institute Inc., Cary, USA). All descriptive data were presented as frequencies for categorical variables and mean (standard deviation [SD]) for continuous variables. Comparisons were made using the chi-square and t-tests, respectively. Time-to-rst switch is presented using Kaplan-Meier curves and 95% con dence limits of the switch rate were calculated assuming a Poisson distribution.

Study sample
For the study period between April 1, 2008 and April 30, 2019, a total of 8,813 patients with a diagnosis of HIV were identi ed; of these, data on 4,293 patients aged ≥ 18 years treated with any ART regimens were extracted and 4,087 patients met the inclusion criteria and were included in the analysis. Among the included patients, 94 (2.3%) patients received a 2-ART on their index date and were classi ed as the 2-ART cohort. There were 3,993 (97.7%) patients who received a 3-ART and were classi ed as the 3-ART cohort. Figure 2 illustrates the ow of patient inclusion. The median duration of follow-up was 712 days and 701 days in the 2-ART cohort and the 3-ART cohort, respectively.

Baseline And Clinical Overview
Of the 94 patients in the 2-ART cohort, the mean age was 54.4 years (SD 12.7 years) and 87.2% were male. In contrast, the 3-ART cohort was signi cantly younger with a mean age of 43.4 years (SD 11.7 years) and included signi cantly more males (93.8%). The proportion of patients having hemophilia was greater in the 2-ART cohort than in the 3-ART cohort (12.8% vs. 2.1%; p < 0.0001). (Table 1).

Co-medications
Of patients in the 2-ART cohort, 92.6% had been prescribed at least one class of co-medication with a mean of 8. respectively. A diagnostic code for AIDS itself was found in 52.1% and 24.8% of the 2-and 3-ART cohorts, respectively. The top 5 most common AIDS-de ning conditions other than AIDS itself was the same for both cohorts: pneumocystis (20.2% and 22.1%), cytomegalovirus disease (17.0% and 10.5%), tuberculosis (6.4% and 6.4%), non-Hodgkins lymphoma (4.3% and 3.2%), and encephalopathy (4.3% and 2.5%).

Comorbidities
Nearly all Charlson comorbid conditions were more prevalent in the 2-ART cohort than in the 3-ART cohort (mean score 6.9 vs. 5.3), with the largest differences found for mild liver disease (

Treatment Patterns
Among patients on a 2-ART, NRTI-sparing regimens comprised the majority of all 2-ARTs (71.3%) (Fig. 3a). Overall, the most common class combinations for this cohort were PI + INSTI (33.0%) and NNRTI + INSTI (31.9%). Among patients treated with a 3-ART, the most common drug class combinations were INSTI + 2NRTI (66.6%), followed by PI + 2NRTI (20.5%) and NNRTI + 2NRTI ( Among patients in the 2-ART cohort, NRTI-sparing regimens were frequently found in all time periods examined (Fig. 3b) ART cohort, and 10.34 switches/100 person-years (95% CI: 9.66-11.06) in the 3-ART cohort; however, no statistical comparisons were made ( Table 2). In the time-to-switch analysis, around 25% of patients in the 2-ART cohort had their rst switch within one year, while only 10% of patients in the 3-ART cohort had their rst switch within one year. As presented in the Kaplan-Meier curve, the higher switch rate in the 2-ART cohort was most evident prior to year 5 of follow-up. However, due to the small patient sample, longterm switch rates were more di cult to evaluate [see Additional le 5]. Of all switches in the 2-ART cohort, 77.4% (24/31) was from 2-ART to 3-ART (

Discussion
To our knowledge, this is the rst study to assess the characteristics of PLWH on a 2-drug regimen, ART treatment patterns, and drug switching in Japan. This retrospective cohort study used over 10 years of hospital-based claim data described the patient characteristics and treatment pattern of 2-ART cohort compared with 3-ART cohort, and explored switching among 2-and 3-ART cohorts in a real-world setting.
Despite the small proportion of 2-ART cohort identi ed, and 2-ART cohort were more experienced at the index date chosen for each cohort, patients on a 2-ART were signi cantly older, received a larger number of co-medications, and had a higher CCI score as well as higher proportion of patients with AIDS-de ning conditions and hemophilia, compared with patients on a 3-ART.
These characteristics suggest that disease status were more severe and complicated in patients on a 2-ART. In addition, hypertension, diabetes, lipid disorders, renal disease and congestive heart failure were more common and signi cantly higher in the 2-ART cohort compared with 3-ART cohort. Similar comorbidities were highlighted in a review of multi-comorbidities in PLWH [20]. The authors also proposed clinical reassessment schedules for at-risk patients, de ned as those who are older than 50 years, have 2 or more comorbidities or receiving polypharmacy (generally considered as 6 or more medications). Another international study also indicated that multi-comorbidities were higher in the 2-ART cohort compared with 3-ART cohort [21]. The authors also found that patients initiation 2-ART were older, had similar regimens compared to our study, and had shorter time-to-discontinuation. [21] Although we did not test any causality, it seems logical that patients who are older, have more co-morbidities, and take more co-medications will likely need a treatment strategy to reduce drug-drug interaction (DDI) or drug adverse events. A 2-ART can be one option for these patients. Also, the higher proportion of renal disease (34.0% vs. 4.8%; p < 0.0001), hemodialysis (6.4% vs. 0.2%; p < 0.0001) and bone disorders (18.1% vs. 6.3%; p < 0.0001) in the 2-ART cohort compared with 3-ART cohort might be a reason for the choice of NRTIsparing regimen. Some studies have demonstrated that greater decrease in renal function and bone mineral density (BMD) were observed in subjects treated with TDF-containing regimens than were observed in subjects treated with TDF-sparing regimens. TDF was well known to cause the proximal tubular and renal toxicities especially in patients with small body weight not only in Japanese but also Vietnamese populations [22][23][24], and they were irreversible after a certain period of TDF use [25]. NRTIsparing regimen comprised the majority of all 2-ARTs in our study, which might suggest that TDF-sparing strategies were adopted to avoid these toxicities. Thus, resorting to non-standard regimens based on patients' more complex and sicker pro les re ect the strategy that we refer to as "negative selection". In contrast, patients who are not as sick as the population identi ed in our study may be also placed on 2-ART regimens an option, especially for naïve patients or treatment-experienced patients with few complications as these could be switches for convenient purposes. We refer to 2-ART regimen selection of this nature as "positive selection". "raltegravir + darunavir/ritonavir" group have more failures in patients with low baseline CD4 cell count [26].
Some debate remains about the usefulness to switch from a 3-ART to a 2-ART regimen. Some experts cite pharmacodynamics concerns about drug plasma level variability with a 2-ART [27], while others cite the bene ts of a reduced pill burden by switching that could lead to better treatment adherence [28]. In the current study, more patients in the 2-ART than the 3-ART cohort switched their ART regimen with a switch rate that was approximately two times higher. Again, this might imply some challenges in managing patients on 2-ART. Two interpretations of these results can be discussed. Either patients were too complex to be managed on a long-term basis, or the 2-ART were not as e cacious or safe compared to the 3-ART. Our study has clearly characterized patients on 2-ART might be di cult to treat regardless of 2 or 3-ART. Another potential reason for a higher switch-rate and earlier switches might be that the 2-ART was not as e cacious or safe as compared SoC 3-ART. Several studies on 2-ART have demonstrated heterogeneous results [29][30][31], some demonstrating inferiority compared to SoC. Recently several 2-ARTs have demonstrated non-inferiority and are now approved in several countries. Since these studies were usually done in the optimal settings of randomized controlled trials with a very controlled patient population, data from real-world settings will need to demonstrate the robustness of these regimens. The reasons for drug switching were not recorded in the MDV database; however, potential reasons include drug-drug interaction, drug resistance or adherence issues. As PLWH in Japan have been noted to have good drug adherence [15], the reasons could be related to negative selection and the unique comorbidities and co-medications of this population.
There are several limitations inherent in database studies that should be noted. The MDV database is not representative of all PLWH in general because hospitals with advanced medical care capabilities are included in the database. Doctors' choices of ART regimens or switching strategies may also differ not only among hospitals in the database but also between these hospitals and HIV-specialized facilities.
There may be misclassi cation of 2-ART and 3-ART cohort due to the assumption that all components of the regimen are prescribed on the same day. It is also not possible to track patients as they move between hospitals. To address this shortfall, sensitivity analysis was performed that required patients to meet criteria for inclusion as "continuously" receiving care and assessed patients from their rst captured switch in the database. There may still be differential follow-up time between 2-ART and 3-ART cohort beyond one year, a potential remaining source of bias when evaluating switch rates. Limitations also exist with the switch analysis for the 3-ART cohort as, due to limited sample size, any patient on a 2antiretrovirals was classi ed into the 2-ART cohort. Therefore, a 3-ART switch could not drop to a 2-ART. In addition, with the 2-ART cohort, the index date is date of initiation of the 2-ART, which could be quite far along in their treatment history and likely when they are older. Finally, lab results, such as viral load and CD4 + counts, were not recorded in the database, only the number of the test was recorded. Therefore, the clinical effects of 2-versus 3-ART could not be assessed.
This study was the rst step in understanding patient pro les and medical needs of those who bene t from 2-ART in Japan. It provides a foundation for future studies to further analyze and explore reasons for switching, as well as identifying those may most bene t from NRTI-sparing regimen. As it is expected that more patients will receive 2-ART by positive selection in the future, research characterizing the bene ts of these regimens is imperative.

Conclusion
In conclusion, patients on 2-ART demonstrated to have complex pro les, are older, and have more comorbidities and co-medications compared to patients on 3-ART. This highlights the negative selection of 2-ART regimens and the need for careful management by physicians to reduce treatment-related toxicities up to now.  Flow chart of patient attrition