Selecting Insulin Analogs for Closed-Loop Control Using Multiplex Pharmacokinetic Profiling: Complete Statistical Analysis

1. Trial at a Glance

This page covers the registry results for Selecting Insulin Analogs for Closed-Loop Control Using Multiplex Pharmacokinetic Profiling (NCT01684943), a non-randomized na study. The primary outcome is For Continuous Insulin Monitoring: Time to Maximum Plasma Insulin and Time to Maximum Continuous Insulin Monitoring Insulin.

2. Clinical Question

3. Trial Design

The registry describes a completed, nonrandomized, single-group study sponsored by Massachusetts General Hospital. The phase is listed as NA. Enrollment was 33. The study began in 2010-07, and primary completion was recorded in 2016-09. Masking was none. The record does not describe a randomized assignment or a parallel control group, so the study should not be read as a like-for-like comparison of independently assigned treatment arms.

The main study used multiplex pharmacokinetic profiling to examine insulin absorption within individual subjects. The brief summary names insulin lispro, insulin aspart and insulin glulisine. The arm description also lists regular human insulin as part of the profiling. The design therefore centers on measurements of pharmacokinetic profiles following injections of insulin types, rather than on a treatment-versus-placebo comparison. Blood samples were drawn at intervals for pharmacokinetic profiling, although the full sampling protocol is not reported in the registry record.

The continuous insulin monitoring work is described as a distinct sub-study, not as another randomized arm of the multiplex study. Some subjects participated in the sub-study, which involved insulin lispro and monitoring its pharmacokinetic profile using blood samples as well as the device. The registry states that only 2 of the 4 experiments conducted under this sub-study protocol produced usable data for analysis, and that both usable experiments are reported.

The adverse-event record uses three participation categories: Multiplex Pharmacokinetic Profiling Only, Continuous Insulin Monitoring Only, and Both Multiplex PK and Continuous Insulin Monitoring. These are safety reporting groups, not randomized comparison groups. The record does not report the allocation process beyond identifying the study as nonrandomized, nor does it report a statistical analysis plan or a sample-size justification.

4. Participant Flow and Accounting

The registry reports how participants moved through each stage of the study. All counts below are exactly as posted; missing counts are not inferred from other stages.

Before assignment: 29 enrolled in Multiplex PK profiling, 9 were ineligible. 21 completed. 3 enrolled in the CIM arm of the trial. 2 completed, 1 was excluded from analysis. 1 subject enrolled in both study phases. They completed the MultiPK profiling experiments, but were excluded from the CIM analysis. Total protocol enrollment was 33.

Overall Study

Reasons for not completing this period:

5. Baseline Characteristics

The table reproduces the baseline characteristics posted in the registry, which describe the participants analysed at baseline and help show whether the randomized groups were comparable.

6. Primary Endpoint

The registry lists the primary outcome as For Continuous Insulin Monitoring: Time to Maximum Plasma Insulin and Time to Maximum Continuous Insulin Monitoring Insulin, measured over 10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 110, 120, 140, 160, 180, 200, 220, 240, 260, 280, 300 minutes after dose.

The primary endpoint concerns two timing measures: time to maximum plasma insulin and time to maximum insulin measured by continuous insulin monitoring. Both are expressed in minutes after dose. The listed measurement times run from 10 through 300 minutes after dose, at the intervals specified in the registry record. The endpoint is intended to describe when each measurement reaches its maximum, allowing the plasma-based profile and device-derived profile to be considered side by side.

As a general statistical concept, time to maximum is a feature of a concentration-time profile. It captures timing rather than the magnitude of the maximum concentration. Comparing timing measures can help assess whether a monitoring signal tracks the timing of the plasma profile, but timing alone does not establish agreement across the full profile or accuracy for every pharmacokinetic parameter.

The endpoint is especially relevant because the study included a preliminary feasibility evaluation of continuous insulin monitoring and its potential use in deriving pharmacokinetic parameters. The registry record identifies the endpoint and its units, but does not report numerical endpoint results, an estimate of the difference between the two timing measures, or a measure of uncertainty. It also does not report the analysis method used to summarize or compare the timing values. Only 2 of the 4 sub-study experiments produced usable data, which limits how much the endpoint can establish about the device's performance across the planned experiments.

7. Why No Re-analysis Is Shown

8. Statistical Interpretation

9. Statistical Methods Explained

The registry describes a nonrandomized, single-group study with within-individual pharmacokinetic profiling. It does not report a statistical analysis plan. The statistical ideas below explain how this kind of endpoint is generally interpreted; they should not be taken as a description of methods that the investigators used unless those methods are explicitly stated in the record.

What does time to maximum measure?

Time to maximum, often abbreviated as Tmax in pharmacokinetic contexts, is the time point at which a measured concentration reaches its observed peak. For the monitoring sub-study, the registry names both time to maximum plasma insulin and time to maximum continuous insulin monitoring insulin. A comparison of these times addresses timing, not the height of the peak, total exposure, or agreement across every measurement in the concentration-time profile.

The listed endpoint timeframe consists of measurements at 10, 20, 30, 40, 50, 60, 70, 80, 90, 100, 110, 120, 140, 160, 180, 200, 220, 240, 260, 280, and 300 minutes after dose. In general, when measurements are taken at discrete times, the observed maximum is identified from the sampled values. The exact timing estimate can depend on the sampling schedule and on how the analysis handles a maximum that falls between measurements. The registry does not report those analysis details.

Why use within-individual profiling?

In general, measuring different insulin profiles within the same individual can make it possible to compare observed patterns while reducing the influence of stable differences between people. This is a paired or repeated-measures idea: each individual's observations can be considered in relation to their own other observations. It does not remove all sources of variation. For example, measurement conditions, timing, or order could still matter, and the registry record does not report enough detail to determine how such factors were handled.

The single-group, nonrandomized classification matters because this study is not a randomized comparison between separate groups. Within-person measurements may address a different question from a parallel-group trial. They can describe relative profiles under the study conditions, but do not by themselves establish what would happen under other conditions or in people not represented in the study.

What would uncertainty communicate?

As a general statistical concept, a confidence interval describes the range of values reasonably compatible with an estimate under the model and assumptions used. A narrow interval indicates greater precision than a wide interval, but neither width nor coverage can be judged without the interval itself and information about the analysis. The registry record does not report a confidence interval or another measure of uncertainty for the primary endpoint. It also does not state a statistical test or model for comparing the two timing measures.

For repeated measurements or small numbers of usable experiments, the choice of summary and analysis can affect interpretation. A mean, median, paired difference, or agreement analysis each answers a somewhat different question. General methods for assessing agreement may consider both systematic differences and variation between paired measurements; a comparison of average timing alone would not necessarily describe individual-level agreement. The registry does not specify which approach was used, so no particular method should be attributed to the investigators.

How should planning illustrations be read?

The record describes a planning illustration based on the observed effect, with a two-sided significance level of 0.05, equal allocation between two arms, and a normal approximation. It states that the effect to detect is a stated share of the observed effect. For time-to-event calculations, the convention is to report the total number of events using the Schoenfeld method; other calculations report participants per arm. These are planning assumptions, not a report of the original study's design or observed power. The study itself is described as nonrandomized and single-group, so a hypothetical equal-allocation two-arm illustration is not a reconstruction of its actual design.

In general, a sample-size illustration answers a conditional planning question: under specified assumptions, what sample size would be associated with a target level of power for a stated effect? It does not show that the completed study had that power, and it does not turn a study's observed estimate into a reliable estimate of the effect to expect in a future study. The registry does not report its own sample-size justification, target power, or design assumptions.

Over-dispersion is a general concern when count data vary more than a simple statistical model expects. If present and ignored, standard errors can be understated and uncertainty can appear too small. The record identifies the primary outcome as a number, but does not report an analysis of over-dispersion or indicate that the endpoint was modeled as a count. The listed outcome is a timing measure, so count-data methods should not be presumed to apply.

10. Multiplicity, Missing Data and Analysis Populations

The record lists 6 outcome measures and gives four secondary outcome descriptions. Those secondary descriptions concern baseline HbA1c in relation to the insulin analog with the most favorable profile for an individual, differences in Tmax between analogs, HbA1c according to use of analogs with different Tmax categories, and baseline hypoglycemia events in relation to the analog with the most favorable profile. The primary endpoint is described separately for continuous insulin monitoring.

When a study evaluates multiple outcomes or comparisons, general statistical practice considers whether the number of analyses increases the chance of at least one apparently notable result. A multiplicity strategy might define a primary analysis, adjust inference across multiple tests, or treat secondary analyses as exploratory. The registry record does not report a multiplicity adjustment, a prespecified hierarchy among outcomes, or a statistical analysis plan. It also reports 0 analyses. As a result, the record does not provide a set of inferential results from which to judge how multiplicity was handled.

For the monitoring sub-study, only 2 of the 4 experiments produced usable data for analysis. The reason for unusable data, any missing measurements within those experiments, and the handling of missingness are not reported in the registry record. The record also does not specify the analysis population for the primary endpoint or for the secondary outcomes. It is therefore not possible to determine whether any analyses used all enrolled participants, only participants with particular measurements, or another defined population.

As a general matter, missing-data methods rely on assumptions about why observations are absent. Without information about the extent, reasons, and handling of missing data, the direction or magnitude of possible bias cannot be assessed. The usable-experiment statement should be read as a limitation of the reported monitoring data, not as evidence about why the other experiments could not be analyzed.

11. Safety Results

The registry reports serious adverse events as affected participants divided by participants at risk.

Deaths

Other (non-serious) adverse events

Participants with at least one non-serious event at or above the registry's 5% reporting threshold.

The registry reports safety counts in three categories: Multiplex Pharmacokinetic Profiling Only, Continuous Insulin Monitoring Only, and Both Multiplex PK and Continuous Insulin Monitoring. In these respective groups, the numbers at risk for serious adverse events were 29, 3, and 1; the serious adverse event counts were 0 in each group. The numbers at risk for other adverse events were 29, 3, and 1; the other adverse event counts were also 0 in each group. Deaths were reported as 0 in each group.

These are registry counts for the listed categories. They do not establish that no adverse events of any kind occurred outside the categories or reporting period captured by the record. The registry information does not give event definitions, observation duration, severity details beyond the serious-versus-other distinction, or a complete account of ascertainment. The count of 0 should therefore be read as no events recorded in that category in the registry, rather than as proof of zero risk. The group counts should not be treated as randomized safety comparisons.

12. What the Result Does, and Does Not, Mean

The record supports a clear description of the study's aims and structure: insulin pharmacokinetic profiles were to be compared within individuals, and a separate preliminary sub-study evaluated continuous insulin monitoring. The primary outcome named for that sub-study is the time to maximum plasma insulin and the time to maximum device-measured insulin. The registry also specifies that only 2 of the 4 experiments produced usable data for analysis.

The record does not provide numerical results for the primary endpoint in the information available for this analysis. It therefore does not show the observed timing values, their difference, or how precisely any comparison was estimated. Without those results, the endpoint description cannot establish whether the device-derived timing was close to plasma timing. Nor does a timing comparison alone establish how well a device captures the full concentration-time profile or whether it derives other pharmacokinetic parameters accurately.

The broader profiling objective is also not a basis for claiming that one insulin analog had a more favorable profile. Although secondary outcomes refer to individual profiles and baseline measures, the corresponding numerical analyses are not reported in the registry record. No conclusion about a treatment's clinical effect or benefit follows from the study description.

More generally, a nonrandomized within-individual pharmacokinetic study can provide information about measured profiles under its study conditions, but it does not have the same interpretation as a randomized trial of clinical outcomes. The registry's completed status indicates the study status, not that all experiments yielded usable data or that every planned result is available. The posted information is most useful for understanding the intended measurements, the distinction between the main study and its sub-study, and the limits on interpreting the reported data.

13. Why This Trial Matters Statistically

This study raises a useful statistical distinction between measuring a biological profile and evaluating a clinical outcome. The multiplex profiling aim concerned the absorption patterns of insulin types within individuals. The continuous monitoring sub-study asked whether a device could measure insulin in a way that might support pharmacokinetic characterization. These questions call for careful attention to the kind of measurement being compared, the timing of observations, and the amount of usable data.

The primary monitoring endpoint focuses on time to maximum, a single feature of a concentration-time profile. General statistical interpretation must distinguish timing from peak magnitude and from agreement across the full profile. A device may be assessed on one feature without that assessment automatically validating every possible use of its measurements. The registry does not provide numerical results or uncertainty estimates for the listed endpoint, so it cannot establish the degree of correspondence between device and plasma timing.

The within-individual design is also statistically informative because it frames comparisons around each participant's measurements rather than separate randomized groups. At the same time, it does not answer every question that a controlled trial could address. The study's nonrandomized, single-group classification should guide interpretation, particularly because the continuous monitoring work was a distinct sub-study rather than a randomized comparator arm.

Finally, the record illustrates why usability and analysis populations matter. Only 2 of the 4 monitoring experiments were usable for analysis. Without reasons for that loss or a fuller account of missing data, the scope and representativeness of the resulting information are difficult to assess. These design and reporting details are central to understanding what a feasibility study can contribute, even when the registry identifies a specific endpoint.

14. Limitations

• The posted summary statistics are the only data these results rest on; patient-level data, full protocols and statistical analysis plans are not available.
• Recalculated results are simple, unadjusted calculations from posted summaries. They do not replace the sponsor's prespecified analysis.

The most direct limitation is that the registry record does not report numerical results for the primary endpoint. It identifies the timing measures and measurement schedule, but gives no observed values, comparison estimate, confidence interval, or statistical test. The record also reports 0 analyses, so the posted information does not permit an independent statistical interpretation of the primary or secondary outcomes.

Usable data were available from only 2 of the 4 experiments conducted under the continuous insulin monitoring sub-study protocol. The reasons the other experiments did not produce usable data are not reported. The registry also does not explain whether there were missing observations within the usable experiments or how missingness was handled. This makes it difficult to judge whether the analyzed experiments represent the intended sub-study.

The study was nonrandomized and single-group, with no masking. Its within-individual profiling objective differs from a randomized comparison between treatment groups. The registry does not report the details of measurement order, possible carryover, or procedures used to address other sources of variation. These should be treated as unreported design details, not assumed to have been controlled.

The record lists 6 outcome measures while providing four secondary outcome descriptions, and it does not report a statistical analysis plan or a multiplicity strategy. It also does not report the analysis populations, sample-size justification, or interim analyses. These omissions constrain interpretation of the intended analysis and the precision or generalizability of any findings. The adverse-event counts are reported by participation category, but the event definitions and observation period are not reported in the registry record.

15. Related Tutorials

16. Related Statistical Calculators

17. Sources

• ClinicalTrials.gov: Selecting Insulin Analogs for Closed-Loop Control Using Multiplex Pharmacokinetic Profiling — NCT01684943.
• Registry record last updated 2019-12-12; status last verified 2019-11.
• Results first posted 2019-12-12.
• Data retrieved from ClinicalTrials.gov on 2026-11-09.
• Group sizes, means, event counts and sponsor-reported p-values are reproduced as posted. Effect estimates, confidence intervals and p-values not attributed to the sponsor are independently calculated from the posted summary statistics.

18. Record Summary

NCT01684943 is a completed, nonrandomized, single-group study sponsored by Massachusetts General Hospital. It enrolled 33 people with type 1 diabetes and had no masking. The study began in 2010-07, and primary completion was recorded in 2016-09. The phase is listed as NA.

The main objective was to compare insulin pharmacokinetics within individuals. A distinct, preliminary continuous insulin monitoring sub-study assessed insulin lispro and the potential for a minimally invasive device to derive pharmacokinetic parameters. Its primary endpoint concerned time to maximum plasma insulin and time to maximum device-measured insulin. Only 2 of the 4 sub-study experiments produced usable data for analysis.

The registry lists 6 outcome measures, reports 0 analyses, and does not provide numerical primary endpoint results in the record information. Safety counts report 0 serious adverse events, 0 other adverse events, and 0 deaths in each of the three listed participation categories. The record does not report a statistical analysis plan, multiplicity approach, missing-data method, or sample-size justification.