This page provides an independent statistical analysis and educational interpretation of publicly reported results. ClinicalTrials.gov provides the official trial registry record.
1. Trial at a Glance
LATTE (NCT03639311) was a phase 2b, open-label, multicenter, rollover study evaluating antiviral activity and safety of long-acting cabotegravir (CAB) plus long-acting rilpivirine (RPV), administered every 2 months, in HIV-positive participants from the LATTE study.
| Feature | LATTE |
|---|---|
| Study acronym | LATTE |
| ClinicalTrials.gov identifier | NCT03639311 |
| Phase | Phase 2 |
| Status | Completed |
| Condition | HIV Infections |
| Design | Open-label, multicenter, rollover, parallel |
| Allocation | Non-randomized |
| Masking | None |
| Primary purpose | Treatment |
| Enrollment | 97 |
| Number of arms | 3 |
| Lead sponsor | ViiV Healthcare |
| Sponsor type | Industry |
| Start date | September 24, 2018 |
| Primary completion date | December 11, 2019 |
2. Clinical Question
The primary statistical question was whether the long-acting cabotegravir plus long-acting rilpivirine regimen administered every 2 months demonstrated antiviral activity at Month 12 in HIV-1-infected, antiretroviral-therapy-experienced participants.
Population
HIV-positive participants from the LATTE study who were antiretroviral therapy experienced.
Intervention
Long-acting cabotegravir plus long-acting rilpivirine administered every 2 months (Q2M).
Comparator regimen
The registry identifies a DTG + RPV regimen within the study's three-arm design.
Primary question
What percentage of participants had HIV-1 RNA ≥50 copies per milliliter at Month 12 according to the FDA Snapshot Algorithm?
3. Trial Design
The non-randomized allocation is an important statistical feature. Unlike a randomized controlled trial, treatment assignment in this study does not provide the same basis for balancing measured and unmeasured baseline factors across treatment groups. Consequently, comparisons between regimens should be interpreted primarily as descriptions of the observed study groups unless the registry specifies an appropriate adjustment strategy.
4. Study Arms and Treatment Structure
Long-acting regimen
- Long-acting cabotegravir (CAB LA)
- Long-acting rilpivirine (RPV LA)
- Administration every 2 months (Q2M)
Reference regimen
- Dolutegravir (DTG)
- Rilpivirine (RPV)
The registry reports three arms overall. The serious-adverse-event data identify the CAB LA + RPV LA Q2M group, the DTG + RPV group, and a long-term follow-up group. The intervention field lists CAB LA, RPV LA, RPV, and DTG as study drugs.
5. Primary Endpoint
| Endpoint | Registry definition | Time frame |
|---|---|---|
| HIV-1 RNA ≥50 copies/mL | Percentage of participants with HIV-1 RNA ≥50 c/mL as per FDA Snapshot Algorithm at Month 12. The HIV-1 RNA ≥50 c/mL per Snapshot algorithm was determined by the last on-treatment HIV-1 RNA measurement within the 12 months analysis visit window. | At Month 12 [Maintenance Phase] |
The endpoint is binary at the participant level: each participant is classified according to whether the prespecified HIV-1 RNA threshold was reached under the FDA Snapshot Algorithm during the specified analysis window. The reported outcome is therefore a percentage rather than a continuous viral-load measurement.
Here, x represents the number of participants classified as having HIV-1 RNA ≥50 c/mL under the Snapshot Algorithm, while n represents the relevant analysis population for the endpoint.
6. Planned Analysis
The registry reports the Month 12 percentage of participants with HIV-1 RNA ≥50 copies/mL according to the FDA Snapshot Algorithm as the primary endpoint. It states that this endpoint was assessed to demonstrate the antiviral activity of the CAB LA + RPV LA Q2M and DTG + RPV regimens.
For an endpoint expressed as the percentage of participants meeting a binary virologic criterion, the fundamental statistical analysis is estimation of the proportion in the relevant analysis population. A confidence interval for the proportion can then quantify sampling uncertainty. When treatment groups are compared, the analysis should account for the non-randomized allocation and the fact that the study was not masked.
Primary estimand
The central descriptive quantity is the percentage of participants with HIV-1 RNA ≥50 c/mL at Month 12 under the registry's Snapshot definition.
Analysis window
The classification uses the last on-treatment HIV-1 RNA measurement within the 12 months analysis visit window.
Typical uncertainty measure
A binomial confidence interval can quantify uncertainty around the observed percentage.
Comparative caution
Because allocation was non-randomized, a difference between observed percentages should not automatically be interpreted as a causal treatment effect.
7. Statistical Methodology
FDA Snapshot Algorithm
The primary endpoint uses the FDA Snapshot Algorithm to classify virologic status at Month 12. The registry specifies that HIV-1 RNA ≥50 copies/mL is determined using the last on-treatment HIV-1 RNA measurement within the 12 months analysis visit window.
This is fundamentally different from simply taking the highest or lowest viral-load measurement during follow-up. The Snapshot approach assigns an endpoint status using a prespecified analysis rule tied to the participant's on-treatment measurement within the defined visit window.
Binary endpoint estimation
Once each participant has a Snapshot classification, the primary endpoint can be expressed as a proportion. If x of n participants meet the HIV-1 RNA ≥50 c/mL criterion, the estimated percentage is based on the ratio of those participants to the relevant analysis population.
The percentage describes the frequency of the prespecified virologic classification in the analyzed population. It is not itself a measure of relative treatment effect.
Confidence intervals for proportions
For a binary endpoint, a confidence interval provides information about the precision of the estimated percentage. With a relatively small study population, interval estimation is particularly useful because the observed percentage can be sensitive to the classification of individual participants.
Non-randomized treatment comparison
The study's allocation is explicitly listed as non-randomized. This changes the interpretation of between-group differences. In a randomized study, randomization is designed to make treatment groups comparable in expectation. In a non-randomized study, observed differences can reflect differences in the populations receiving each regimen as well as treatment itself.
Consequently, a simple difference in Snapshot percentages is descriptive unless the statistical analysis specifies an adjustment strategy that addresses relevant sources of confounding.
Missing and off-treatment observations
The registry's endpoint definition specifically refers to the last on-treatment HIV-1 RNA measurement within the 12 months analysis visit window. That rule is important because treatment status and the timing of the last measurement are part of the endpoint definition rather than incidental details.
8. Statistical Methods Explained
Why is the primary endpoint expressed as a percentage?
The endpoint asks whether HIV-1 RNA is at or above a prespecified threshold. That creates a binary outcome for each participant. Summarizing the number meeting the criterion as a percentage provides an interpretable measure of the frequency of the virologic outcome.
What does the Snapshot Algorithm add?
The Snapshot Algorithm establishes a consistent rule for assigning virologic status. Rather than allowing investigators to choose whichever measurement is most favorable or informative after examining the data, the registry specifies that the last on-treatment HIV-1 RNA measurement within the 12 months analysis visit window determines the HIV-1 RNA ≥50 c/mL classification.
Why does non-randomized allocation matter?
Without randomization, treatment groups can differ systematically before treatment begins. Those differences can influence the observed Month 12 virologic percentages. Therefore, a between-group percentage difference cannot automatically be interpreted as the causal effect of CAB LA + RPV LA Q2M compared with DTG + RPV.
Why use a confidence interval for a percentage?
An observed percentage is an estimate from a finite sample. The confidence interval communicates statistical uncertainty around that estimate. A narrow interval indicates greater precision than a wide interval under the specified statistical framework, but it does not describe the range of individual patient outcomes.
Why is the Month 12 window important?
The endpoint is not simply "viral load at approximately one year." The registry defines a specific analysis visit window and specifies that the last on-treatment measurement within that window determines the endpoint classification. Changing that rule could change which observation is used and therefore could change the endpoint.
Does a percentage difference establish treatment superiority?
No. A numerical difference between two percentages describes the observed groups. Because the study is non-randomized, establishing a causal treatment effect would require appropriate consideration of the factors that influence treatment assignment and outcome, in addition to estimation of the endpoint itself.
9. Safety Results
The registry reports serious adverse events by study arm. These figures describe the number of participants affected relative to the number at risk in each reported group.
| Group | Participants with serious adverse events | At risk |
|---|---|---|
| CAB LA + RPV LA Q2M | 10 | 90 |
| DTG + RPV | 0 | 7 |
| LTFU: CAB LA + RPV LA Q2M Group | 0 | 6 |
The denominators are important when interpreting these counts. A count of zero in a group with 7 participants or 6 participants carries considerably less information about the underlying event frequency than a zero observed in a much larger population.
The registry reports serious adverse events as affected participants divided by participants at risk. These are descriptive safety counts. They should not be interpreted as evidence that the underlying serious-adverse-event probabilities are identical across groups, particularly when the denominators are small and treatment allocation is non-randomized.
10. Trial Timeline
Study start
The LATTE study began enrollment and follow-up under its phase 2, open-label, multicenter rollover design.
Primary endpoint assessment
The primary endpoint assesses the percentage of participants with HIV-1 RNA ≥50 copies/mL according to the FDA Snapshot Algorithm during the Month 12 maintenance-phase analysis window.
Primary completion
The registry lists December 11, 2019 as the primary completion date.
11. Interpreting the Primary Endpoint
The primary endpoint measures the percentage of participants classified as having HIV-1 RNA ≥50 copies/mL under the FDA Snapshot Algorithm at Month 12. It is a prespecified virologic classification, not a continuous measure of the magnitude of viral-load change.
The endpoint does not by itself quantify the size of viral-load reduction, the durability of virologic suppression beyond the specified analysis window, or an individual participant's long-term clinical outcome.
A percentage always depends on the analysis population used as its denominator. In this study, the endpoint definition additionally specifies the use of the last on-treatment HIV-1 RNA measurement within the 12 months analysis visit window.
Even a large observed difference between treatment-group percentages would not automatically establish that the treatment caused the difference. The study's non-randomized allocation means that treatment-group composition and other differences can contribute to the observed outcome.
12. Analysis of the Study's Statistical Structure
| Statistical feature | LATTE | Interpretation |
|---|---|---|
| Allocation | Non-randomized | Between-group differences require caution because randomization does not provide the basis for causal comparability. |
| Masking | None | The study was open-label, so participants and investigators were not masked to treatment assignment. |
| Design model | Parallel | Study groups are followed as parallel treatment groups rather than through a crossover structure. |
| Primary endpoint | Binary virologic classification | The principal result is naturally expressed as a percentage of participants meeting the HIV-1 RNA ≥50 c/mL criterion. |
| Endpoint time frame | Month 12 [Maintenance Phase] | The analysis is tied to a prespecified follow-up window rather than an unrestricted time point. |
| Snapshot rule | Last on-treatment HIV-1 RNA measurement within the 12 months analysis visit window | The endpoint depends on a defined measurement-selection rule. |
| Formal statistical analyses posted | None | The registry does not provide posted formal statistical estimates, confidence intervals, or p-values for the primary endpoint. |
13. What Can and Cannot Be Concluded From the Registry Results
The registry establishes that results are posted and identifies the primary endpoint and its definition. It also reports serious adverse-event counts by arm. However, the posted statistical-analysis field contains no formal statistical analyses for the primary endpoint.
That distinction matters because a posted endpoint result and a formal inferential analysis are different levels of statistical information. A percentage can describe what occurred in a study population, whereas a confidence interval quantifies uncertainty around the estimate and a hypothesis test addresses a prespecified inferential question.
For LATTE, the non-randomized allocation adds another layer of interpretation. Even if treatment-group percentages are different, the statistical meaning of that difference depends on the comparability of the groups and on the analysis strategy used to address potential confounding.
14. Important Limitations and Interpretation Issues
- Non-randomized allocation: treatment groups are not created by random assignment, so observed between-group differences may reflect differences in participant characteristics or other factors as well as treatment.
- Open-label design: masking was listed as none. Knowledge of treatment assignment can influence aspects of study conduct and assessment, although the primary virologic endpoint uses a defined laboratory-based Snapshot rule.
- Limited enrollment: the study enrolled 97 participants, so percentage estimates can be sensitive to individual participant outcomes.
- Small safety denominators: the registry reports 7 participants at risk in the DTG + RPV group and 6 in the long-term follow-up group for the listed serious-adverse-event summaries. Zero observed events in such small groups do not establish a zero underlying event probability.
- Endpoint classification: the primary endpoint uses the last on-treatment HIV-1 RNA measurement within a defined analysis window. The classification therefore depends on both the virologic measurement and the prespecified Snapshot rule.
- Formal inference: no formal statistical analyses are posted in the registry's statistical-analysis field for the primary endpoint. Consequently, the registry record does not provide a formal confidence interval or p-value for that endpoint.
- Rollover population: participants came from the LATTE study, so the study population represents a rollover cohort rather than a newly assembled randomized treatment population.
15. Why This Trial Matters Statistically
LATTE is a useful statistical teaching case because it illustrates how endpoint definition and study design affect interpretation. The primary outcome is simple to state—a percentage of participants with HIV-1 RNA ≥50 copies/mL—but the correct interpretation depends on the Snapshot classification rule, the analysis window, the treatment exposure status, and the non-randomized study design.
| Concept | How it appears in LATTE |
|---|---|
| Binary endpoint | Participants are classified according to whether HIV-1 RNA is ≥50 copies/mL. |
| Percentage estimation | The primary endpoint is reported as the percentage of participants meeting the virologic criterion. |
| Snapshot analysis | The last on-treatment HIV-1 RNA measurement within the 12 months analysis visit window determines the classification. |
| Time-window definition | The endpoint is evaluated at Month 12 in the maintenance phase. |
| Non-randomized allocation | Treatment-group comparisons do not have the causal protection provided by randomization. |
| Open-label design | Masking is listed as none. |
| Small safety groups | Serious-adverse-event denominators vary substantially among the reported groups. |
| Rollover design | The participants originated from the LATTE study. |
| Confidence intervals | For a binary percentage, interval estimation is the natural way to describe statistical precision. |
| Formal hypothesis testing | No formal statistical analyses are posted for the primary endpoint in the registry. |
16. Clinical Biostats Interpretation
Endpoint interpretation
The primary endpoint is a clinically interpretable virologic threshold: the percentage of participants with HIV-1 RNA ≥50 copies/mL at Month 12 under the Snapshot Algorithm.
Statistical interpretation
The natural statistical summary is a binomial proportion with an uncertainty interval. The registry does not post a formal inferential analysis for this endpoint.
Design interpretation
The non-randomized allocation means that observed differences between treatment groups should not automatically be interpreted as causal treatment effects.
Safety interpretation
Serious-adverse-event counts should be interpreted together with their denominators, particularly where the reported groups contain only 7 or 6 participants at risk.
The most important statistical lesson is that an endpoint can be precisely defined without making the underlying treatment comparison randomized. LATTE combines a clearly specified virologic classification with a non-randomized rollover design. Those two features answer different questions: the Snapshot Algorithm standardizes how the virologic outcome is classified, while the allocation method determines how confidently differences between treatment groups can be interpreted as treatment effects.
17. Sources
- ClinicalTrials.gov: LATTE (NCT03639311).
Continue through the Clinical Biostats clinical-trial library
Explore additional clinical trial statistical analyses and educational material on study design, endpoints, and statistical interpretation.
18. Record Summary
LATTE is a phase 2, open-label, multicenter, non-randomized rollover study of long-acting cabotegravir plus long-acting rilpivirine administered every 2 months in HIV-positive participants from the LATTE study. Its primary endpoint is the percentage of participants with HIV-1 RNA ≥50 copies/mL according to the FDA Snapshot Algorithm at Month 12 during the maintenance phase, using the last on-treatment HIV-1 RNA measurement within the 12 months analysis visit window.
Statistically, the study is most useful for understanding the distinction between endpoint definition and causal treatment comparison. The Snapshot rule provides a standardized binary virologic outcome, while the non-randomized allocation limits how a difference between treatment groups can be interpreted. The registry also reports serious-adverse-event counts by arm, with 10 of 90 participants in the CAB LA + RPV LA Q2M group, 0 of 7 in the DTG + RPV group, and 0 of 6 in the listed long-term follow-up group.