Cohort Characteristics

The flowchart of the study design is shown in Fig. 1.

Fig. 1

Flowchart of the study design. LA-CAB/RPV  long-acting antiretroviral therapy with cabotegravir and rilpivirine (created in BioRender, Dannenberg, C. (2025) https://BioRender.com/7p08hou)

A total of 109 individuals requested initiation or continuation of LA-CAB/RPV injection therapy at our center during the observation period. Three participants discontinued during the oral lead-in phase, reverting to their previous oral ART regimen owing to sleep disturbance (n = 2) or an unspecified reason (n = 1). Consequently, 106 individuals received at least one LA-CAB/RPV injection. Four participants discontinued after the first injection owing to injection site pain and resumed oral ART, leading to their exclusion from subsequent analysis. Thus, 102 PLWH received at least two LA-CAB/RPV injections during the observation time. This group included (a) 77 injection-naïve individuals, (b) 20 injection-experienced individuals, with a median duration of 54.9 months on LA-CAB/RPV during previous clinical trials in their medical history outside this observation period in a real-world setting (range: 32.2–84.8 months; 91.3 patient-years accumulated; Fig. 2). In addition, we included five off-label injection-naïve individuals with baseline VL ≥ 50 copies/mL. These five individuals did not meet the LA-CAB/RPV treatment guideline criteria. They were excluded from the primary adherence and efficacy analysis but assessed separately for adherence, safety, and outcomes (Section: Outcomes of Injection-Naïve Individuals Treated Off-Label).

Fig. 2

Individual LA-CAB/RPV therapy duration in injection-experienced individuals (cohort (b)). The light grey background highlights the observation period. Yellow bars represent therapy duration during prior clinical trials, while orange bars indicate continuation of treatment to routine clinical care within the observation period. The transition from clinical trial participation to routine care occurred after the end of the previous study, not at the beginning of the observation period. Participants ID-HH 7, ID-HH89, and ID-HH 103 were lost to follow-up owing to relocation (created in BioRender, Dannenberg, C. (2025) https://BioRender.com/g4dobmp)

Demographic and clinical characteristics of the study participants are detailed in Table 1.

Table 1 Demographic and clinical characteristics of the study population

The median age of all 102 participants was 49 (range: 22–78) years, with 20% identifying as female. The majority of the individuals were white, primarily from Europe (80%), with 69% born in Germany. A total of 78% transitioned from an oral INSTI-based ART regimen to long-acting injectables. Participants in cohort (a) had a median of 9 years (0–26) on oral ART before switching to LA-CAB/RPV, whereas those in cohort (b) had a median of 2 years (0–11) before entering clinical trials.

Detailed Adherence Analysis on LA-CAB/RPV in Participants Treated In-Label

We conducted a detailed analysis of the adherence patterns of individual injections across both cohorts under real-world conditions, comparing those who initiated LA-CAB/RPV during this study period with those who transitioned into routine care after an extended period of injection therapy in the setting of a registration trial (see above and Table 2).

Table 2 Detailed overview of injection timing

Over the 44-month observation period, a total of 1690 LA-CAB/RPV injections were administered to 97 in-label participants (cohort (a), n = 77; cohort (b), n = 20). Overall, 84% of injections (1417/1690) were administered on time within the interval of 56 ± 7 days (Fig. 3).

Fig. 3

Adherence to the time schedule among in-label participants. Injection timing adherence in cohort (a) and cohort (b) across 1690 administered injections, shown in dark grey box plots. The grey-shaded area represents the tolerance window (± 7 days from the scheduled injection date); injections to the left were administered early, while those to the right were delayed. The purple box plots represent the percentage of viral load increases corresponding to each adherence category. Percentages were calculated on the basis of the number of injections within each adherence group for which a blood draw was available; n number, VL viral load (created in BioRender, Dannenberg, C. (2025) https://BioRender.com/fzwg2n9)

In cohort (a), participants received a median of 21 injections (2–24). Delayed injections occurred in 79% of participants (61/77), and 30% (23/77) experienced more than three delays. Among delayed injections, the median delay was 11(8–69) days, with 97% (59/61) classified as short delays (8–14 days). Delay frequency slightly increased over time: 15.2% in months 0–15 (94/620), 16.9% in months 16–27 (57/337), and 18.1% in months 28–39 (45/248).

In cohort (b), the median number of injections was 19 (7–21). Delays occurred in 95% of participants (19/20), and 20% (4/20) experienced more than three delays. The median delay was 10 (8–112) days, with 95% (18/19) limited to short delays. Delay rates also rose slightly over time: 9.9% (months 0–15), 15.9% (months 16–27), and 16.6% (months 28–39). One individual (ID HH-92) experienced a prolonged delay of 112 days between the final trial dose and the first outpatient injection; this case did not result in CVF.

At the end of the observation period, the participants who were treated in-label had accumulated a total of 348.7 patient-years on LA-CAB/RPV—205.1 patient-years in cohort (a) (median: 37.7 months; range: 0.9–43.4) and 143.6 patient-years in cohort (b) (median: 81.2 months; range: 58.9–121.4), including 52.4 patient-years during our study period.

A total of 14% of participants (14/97) discontinued LA-CAB/RPV and returned to oral ART—all from cohort (a) (18%, 14/77); no discontinuations were observed in cohort (b). An additional 6% (6/97) relocated (cohort (a), 4%, 3/77; cohort (b), 15%, 3/20), and 1% (1/97) was lost to follow-up due to an unknown reason (cohort (a), 1%, 1/77; cohort (b), 0%). Notably, 29% of all discontinuations (4/14) were observed within the first three injections and 50% (7/14) within the first 9 months of injectable therapy. Injection site pain was cited as the leading cause of discontinuation in 36% of cases (5/14), while logistical restraints were reported by 29% (4/14; Table 2).

Treatment Efficacy in Participants Treated In-Label

Since highly detailed data were available, we aimed to create a granular analysis of each participant’s longitudinal adherence patterns and virological efficacy (Fig. 4a, b). VL increases were observed in 20% of individuals (19/97) treated in-label throughout the study (cohort (a) n = 15; cohort (b) n = 4), including 14 participants with a single viral blip, two with recurrent blips, two with LLV, and one case of CVF (ID HH-84), which was preceded by two recurrent blips. Overall, only 1.5% of blood draws during the observation period showed VL ≥ 50 copies/mL (25/1690), including 2.4% following delayed injections (6/247) and 1.3% after on-time injections (18/1417) (Fig. 3).

Fig. 4

a Adherence and virological outcomes in injection-naïve individuals (cohort a) (created in BioRender, Dannenberg, C. (2025) https://BioRender.com/eawxe5w); b Adherence and virological outcomes in injection-experienced individuals (cohort b). Adherence and virological outcomes of LA-CAB/RPV treatment over the timeline (x-axis) for each individual (y-axis). Adherence is visualized by length of the intervals between injections, represented as boxes along each participant’s timeline. Delays beyond the scheduled injection dates are annotated with the corresponding number of days overdue. To ensure visibility of initial VL data, the width of the first interval block was standardized. Virological outcomes are visualized by different colors: light green: undetectable VL (< 13.2 copies/mL); green: VL < 50 copies/mL; red: VL 51–200 copies/mL; and dark red: VL > 200 copies/mL. Follow-up visits at 12–15 months, 24–27 months, and 36–39 months are marked by vertical lines. *Discontinuations, CVF, LTFU, and relocations were annotated; n number, CVF confirmed virological failure, LA-CAB/RPV long-acting antiretroviral therapy with cabotegravir and rilpivirine, LTFU lost to follow-ups, VL viral load (created in BioRender, Dannenberg, C. (2025) https://BioRender.com/y5txl4d)

In cohort (a), virological data were analyzed for 67 individuals at the 12–15-month follow-up window, 56 individuals at 24–27 months, and 44 individuals at 36–39 months (Fig. 5).

Fig. 5

Virological outcomes at follow-up time windows among in-label participants. Long-term virological outcomes of all available in-label participants at defined follow-up time windows, stratified by HIV-1 VL < 50 copies/mL, VL ≥ 50 copies/mL, and undetectable VL (< 13.2 copies/mL). The number of participants varies due to loss to follow-up (e.g., discontinuation, relocation, or other reasons) or/and the end of the observation period. In cohort (a), data were available up to the 36–39-month follow-up window. As the observation period ended after 44 months, no data from the cohort (a) were available for the 48–51-month window or beyond. In cohort (b), injection-experienced participants transitioned to routine care at different time points. The earliest transitions occurred after 32 months of LA-CAB/RPV therapy and involved two individuals. By the 48–51-month follow-up, n = 9 participants had entered the observational phase, increasing to n = 13 individuals at 72–75 months. Thereafter, participant numbers declined due to the end of the observation period. Notably, n = 3 individuals remained on LA-CAB/RPV therapy for a full 10 years and consistently maintained viral loads < 50 copies/mL. n number, HIV-1 human immunodeficiency virus type 1, LA-CAB/RPV long-acting antiretroviral therapy with cabotegravir and rilpivirine, VL viral load (created in BioRender, Dannenberg, C. (2025) https://BioRender.com/e53rcfh)

A total of 99% of evaluable participants at the 12–15-month follow-up window achieved virological suppression (66/67) and 74% undetectable VL (49/67). During this period (1–15 months), 13% of all individuals (10/77) experienced at least one VL increase ≥ 50 copies/mL. This included eight single blips (80%, 8/10), one individual with two blips (10%, 1/10), and one case of LLV (10%, 1/10). Notably, 67% of all VL increases in cohort (a) occurred within the first 15 months (10/15). At the 24–27 and 36–39-month follow-up windows, all available individuals maintained viral suppression, with 75% (42/56) and 77% (34/44) having undetectable VL. However, four individuals experienced a single blip between months 16 and 39, one participant exhibited a recurrent blip, and one case of CVF occurred after 40 months of injection therapy (ID HH-84). This case is analyzed in detail in a separate section (see section CVF).

At the final blood draw of the observation period, 98% of the 59 individuals in the cohort (a) who remained on LA-CAB/RPV achieved VL < 50 copies/mL (58/59), and 73% had undetectable viral load (43/59). Excluding blips, 97% maintained viral suppression throughout the study (75/77), with one CVF and one LLV.

For cohort (b), the total time on treatment was calculated, and virological outcomes were assessed for 9 individuals at months 48–51, 13 at months 72–75, four at months 96–99, and three at months 120–123. All individuals maintained a VL < 50 copies/mL in the defined follow-up windows (Fig. 5). However, four individuals experienced at least one VL increase over the observation period, including three single blips and one case of LLV. At the time of the VL increase, all of them had been on LA-CAB/RPV therapy for at least 4–5 years. At the last blood draw of the observation period in cohort (b), 100% of the 17 individuals who remained on LA-CAB/RPV therapy achieved a VL of less than 50 copies/mL, and 76% (13/17) had an undetectable viral load. Excluding blips, 95% maintained viral suppression throughout the study (19/20). In total, 64% (144/225) of all HIV-1 VL measurements were undetectable (< 13.2 copies/mL).

To evaluate the impact of very low-level viremia, we compared participants of cohort (a) with undetectable VL (< 13.2 copies/mL) with those with suppressed but detectable VL at baseline (< 50 copies/mL). Overall, 65% (50/77) initiated LA-CAB/RPV therapy with an undetectable VL. During follow-up, 18% (9/50) of this group experienced VL increases ≥ 50 copies/mL, all of which were isolated blips. In contrast, among participants with suppressed VL at baseline, 22% (6/27) showed VL increases, including two single blips, three recurrent blips (one progressing to CVF), and one LLV.

Outcomes of Injection-Naïve Individuals Treated Off-Label

Five injection-naïve individuals initiated LA-CAB/RPV off-label with a baseline VL ≥ 50 copies/mL (median: 87 copies/mL; range: 55–1830; n = 4 VL < 200 copies/mL). The median therapy duration was 22.6 months (0.9–40.0), corresponding to 9 patient-years. Three of five participants experienced at least one delay (60%), and one had more than three delays. Most delays occurred between 8 and 14 days (71%, 5/7). Delays > 56 days were observed in two individuals, resulting in one CVF with a delay of 959 days (Table 3).

Table 3 Participants with CVF

At the end of the observation period, three participants remained on injectable treatment (60%). Among them, two participants (ID HH-56, baseline VL 55 copies/mL; ID HH-99, baseline VL 87 copies/mL) maintained virological suppression throughout 37 and 40 months on LA-CAB/RPV, respectively. The third individual (ID HH-101) switched to LA-CAB/RPV therapy with a baseline VL of 1830 copies/mL but never achieved virological suppression during 22 months of treatment. To exclude pharmacokinetic failure, plasma drug concentrations were evaluated at the time of the fourth injection, confirming therapeutic levels: cabotegravir at 1840 µg/L (expected trough levels for 8-week dosing: 800–3000 µg/L) and rilpivirine at 78 µg/L (expected: 40–120 µg/L). On the basis of multidisciplinary expert consensus, the participant was classified as having non-suppressive viremia (NSV). In alignment with the individual’s preference to reduce oral pill burden, LA-CAB/RPV therapy was continued.

Two individuals treated off-label (ID HH-43 and ID HH-78) discontinued therapy owing to virological failure (section CVF).

Confirmed Virological Failure (CVF)

Three cases of CVF were observed during the study period (ID HH-43, ID HH-78, and ID HH-84), two of which occurred in individuals who initiated LA-CAB/RPV with a baseline VL ≥ 50 copies/mL (off-label). All data are summarized in Table 3, with each case of CVF illustrated in Supplementary Fig. S1a–c (Electronic Supplementary Material). ID HH-43 (Supplementary Fig. S1a) initiated LA-CAB/RPV six months after the HIV diagnosis with a baseline VL of 117 copies/mL, prior to achieving full viral suppression. Off-label LA-CAB/RPV initiation was motivated by high psychosocial burden related to stigma. The patient had been hospitalized owing to Pneumocystis jirovecii pneumonia (PJP) and met the Centers for Disease Control and Prevention (CDC) stage C3 criteria (VL peak 615,000 copies/mL, CD4+ nadir 5 cells/μL). A resistance test conducted four months prior to injection therapy indicated subtype B and low-level resistance to RPV. After only two on-time LA-CAB/RPV injections, VL increased to 14,000 copies/mL. Repeated resistance testing revealed broad NNRTI resistance and INSTI resistance. Consequently, the participant was switched to a PI-based oral ART (darunavir (DRV)/tenofovir alafenamide (TAF)/emtricitabine (FTC)) and achieved viral suppression after six months.

ID HH-78 (Supplementary Fig. S1b) initiated LA-CAB/RPV with a baseline VL of 69 copies/mL, following 2.5 years of viral suppression on oral ART (bictegravir (BIC)/TAF/FTC). At HIV diagnosis, VL peak was 27,500 copies/mL, and CD4+ nadir was 568 cells/μL. Baseline resistance testing revealed no major mutation. Under LA-CAB/RPV, VL was suppressed from the fourth injection onward. Apart from one 8-day injection delay, all injections were on time. After the sixth injection, the participant was lost to follow-up for 959 days owing to a psychosocial crisis. Upon return, VL had increased to 16,000 copies/mL, with resistance testing revealing broad NNRTI resistance. The individual resumed INSTI-based oral ART (BIC/TAF/FTC) and successfully suppressed after one month.

ID HH-84 (Supplementary Fig. S1c) initiated LA-CAB/RPV in-label with oral lead-in and a baseline VL of 33 copies/mL. HIV had been diagnosed 10 years earlier with PJP and Kaposi’s sarcoma (CDC C3; peak VL 500,000 copies/mL; CD4+ nadir 5 cells/μL). Prior ART included NNRTI-based and INSTI/protease (PI) dual regimen, most recently BIC/TAF/FTC. Two resistance tests showed no major resistance. Over 40 months, the participant received 22 injections, experiencing two blips and two minor delays. At 40 months, VL increased to 10,100 copies/mL. Resistance testing revealed extensive resistance mutations to NNRTIs and INSTIs. The individual was switched to a PI-based ART and re-suppressed within two months.

Side Effects and Adverse Events

Local injection site reactions were reported in 63% of participants (64/102), with pain being the most frequently observed symptom (92%, 59/64; Table 4). The most painful reactions occurred during the initial treatment phase, with the first three injections accounting for 62.0% of all reported painful injections (69/112). Following the third injection, the incidence of reported pain decreased substantially; only 17% of affected individuals (10/59) continued to report pain. Notably, injection site pain contributed to 36% (5/14) of all treatment discontinuations observed among in-label treated participants (Table 2).

Systemic side effects were reported in 42 participants (41%, 42/102) at some point during the observation period. Most common were neuropsychiatric side effects (60%, 25/42, all treated in-label), with depressive episodes being the predominant neuropsychiatric symptom (64%, 16/25). Although formal grading of severity was unavailable, most events were described as mild and did not result in treatment discontinuation or hospital admission. Notably, 75% of individuals with depressive symptoms (12/16) reported a history of prior depressive episodes. Other systemic side effects included metabolic disturbances (36%, 15/42) and cardiovascular problems, specifically hypertension (14%, 6/42).

Routine laboratory parameters for immunological, metabolic, hepatic, and renal function were longitudinally evaluated in injection-naïve (a) and treatment-experienced participants (b) and are presented in Supplementary Table S1a and S1b (Electronic Supplementary Material). No relevant abnormalities or notable differences were observed between the cohorts. All metabolic, hepatic, and renal parameters (including cholesterol, triglycerides, LDL, GPT, lipase, and creatinine) remained stable throughout the observation period, with no indication of long-term changes.

The median baseline CD4+ T cell count was 630 cells/µL in cohort (a) (range: 163–1251, n = 73) and 617 cells/µL in cohort (b) (range: 150–1306, n = 19). Follow-up demonstrated stable immune profiles: cohort (a) had a median of 705 cells/µL (110–1307) at 36–39 months, while cohort (b) maintained medians of 712 cells/µL (233–1241) at 72–75 months and 595 cells/µL (508–886) at 120–123 months.