Enrichment Strategies for Medical Device Clinical Trials: Patient Selection Guide

Designing a pivotal Investigational Device Exemption (IDE) clinical investigation or Premarket Approval (PMA) study for an innovative Class III medical device is one of the most resource-intensive undertakings in medical technology development. Unlike pharmaceutical trials, which can frequently scale to thousands or tens of thousands of participants across global trial networks to detect modest treatment effects, medical device investigations are constrained by physical realities: invasive procedural risks, specialized operator learning curves, substantial per-patient device and hospital costs, and the practical impossibility of enrolling massive cohorts.

When a device trial enrolls an unselected, "all-comers" patient population, the true clinical effect of the technology is frequently diluted. Patients with mild disease may experience too few clinical events to demonstrate statistical superiority or non-inferiority, while patients with anatomical mismatches or non-responsive pathophysiological subtypes dilute the treatment effect. The result is an underpowered trial, ballooning study timelines, inflated sample sizes, or catastrophic trial failure.

Clinical trial enrichment—the prospective, deliberate selection of a study population in which the hypothesized treatment effect of an investigational medical device is more readily detectable than in an unselected population—provides the methodological solution. However, enrichment is a strategic double-edged sword. Every inclusion criterion, imaging cutoff, or severity gate added to an IDE protocol to maximize statistical power simultaneously defines and constrains the device's eventual FDA-approved indication for use, commercial addressable market, and clinical generalizability.

This guide provides medical device regulatory affairs executives, biostatisticians, Chief Medical Officers, and clinical operations leaders with an authoritative decision framework for designing, implementing, and defending prospective enrichment strategies in medical device clinical trials.

---

Executive Summary & Core Decision Framework

Scenario & Direct Answer

Sponsor Scenario: We are designing a pivotal IDE investigation for a novel Class III therapeutic implant (e.g., a transcatheter structural heart system, active neurostimulator, or bioresorbable scaffold). Preliminary early feasibility study (EFS) data show promising efficacy, but high procedural costs and significant patient response heterogeneity make an all-comers trial financially and operationally unfeasible. How do we choose between prognostic, predictive, and variability-reduction enrichment strategies, which device-specific screening gates will FDA CDRH accept, and how do we ensure our narrow enrollment criteria survive scrutiny under FDA Diversity Action Plan (DAP) mandates?

Direct Answer: Device sponsors should prospectively select enrichment strategies based on the primary clinical failure mode of the trial. If the primary risk is an inadequate number of clinical events, deploy prognostic enrichment using validated severity thresholds (e.g., quantitative imaging biomarkers, functional class, or core-lab hemodynamic criteria) to elevate control-arm event rates, which can reduce required sample size by 3x to 4x. If the primary risk is variable therapeutic response, deploy predictive enrichment using physiological or anatomical gates (e.g., ECG morphology, endoscopic phenotyping, or quantitative defect sizing) where the device mechanism directly matches patient biology. For implantable neuromodulation, utilize device-unique run-in screening trials (such as temporary percutaneous leads). To survive FDA CDRH review and FDORA Section 3601 DAP requirements, sponsors must justify all enrichment criteria on objective pathophysiological and anatomical mechanisms rather than demographic proxies, pre-specify core-lab adjudication charters, and structure a phased post-market indication expansion pathway from study initiation.

Core Decision Matrix: Enrichment Archetypes in Medical Devices

The following matrix compares the four primary enrichment strategies utilized in medical device clinical investigations, detailing their data requirements, sample size impact, regulatory scrutiny, and downstream indication trade-offs.

Enrichment Archetype	Clinical Mechanism & Definition	Device-Specific Selection Tool	Statistical & Sample Size Impact	Impact on FDA Indication & Market	CDRH Review Considerations
Prognostic Enrichment	Enrolls patients with a high baseline probability of experiencing the primary clinical endpoint (e.g., death, stroke, heart failure hospitalization).	Core-lab quantitative echo, CT scoring, baseline biomarker cutoffs, functional classification (e.g., NYHA III/IV).	Elevates control-arm event rates; reduces required sample size $N$ by 50% to 75% for event-driven time-to-event endpoints.	Restricts initial labeled indication to high-risk or severe sub-populations.	High acceptance if severity criteria reflect standard clinical practice guidelines; requires justification of generalizability.
Predictive Enrichment	Enrolls patients who possess an anatomical, physiological, or biological marker predicting differential response to the device mechanism.	ECG morphology (e.g., LBBB), Drug-Induced Sleep Endoscopy (DISE), structural geometry, Companion Diagnostics (CDx).	Maximizes absolute effect size ($\delta$); prevents treatment effect dilution by non-responsive anatomical phenotypes.	Indication restricted strictly to the biomarker-positive or anatomically compatible sub-population.	Scrutinized for diagnostic accuracy; if an IVD or software algorithm is used as a selector, may require co-developed CDx or IDE approval.
Variability Reduction	Enrolls a homogeneous population to minimize baseline outcome noise, spontaneous symptom resolution, or medication adjustments.	Run-in medication optimization phases, baseline stability periods, centralized core-lab scoring.	Decreases outcome variance ($\sigma^2$); sharpens precision of continuous functional endpoints (e.g., 6MWD, PROs, KCCQ).	Minor impact on indication; may restrict label if run-in requires strict intolerance or failure of prior lines.	Welcomed by CDRH reviewers; background medical therapy must be rigorously standardized across both treatment and control arms.
Device Run-In / Trial Period	Evaluates acute or sub-chronic therapeutic response using a temporary, externalized, or minimally invasive device before permanent implantation.	3- to 7-day percutaneous neurostimulation screening trials (e.g., SCS $\ge 50%$ pain relief threshold).	Creates a pre-selected responder cohort; substantially boosts permanent implant responder rates in RCTs.	Approved indication typically mandates trial-period success prior to permanent implant reimbursement.	Requires transparent reporting of screening-phase attrition and screen-failure characteristics to assess clinical validity.

---

What Are Prognostic, Predictive, and Variability-Reduction Enrichment Strategies?

To structure an effective clinical protocol, biostatisticians and regulatory affairs professionals must understand how classical clinical trial enrichment principles apply to physical, mechanical, and electrical medical technologies.

+----------------------------------------------------------------------------------------------------+
|                                 MEDICAL DEVICE ENRICHMENT SPECTRUM                                 |
+------------------------------------+----------------------------------+----------------------------+
|        PROGNOSTIC ENRICHMENT       |      PREDICTIVE ENRICHMENT       |   VARIABILITY REDUCTION    |
|   (Elevating Baseline Event Rate)  |   (Isolating Device Responders)  |   (Minimizing Noise & Var) |
+------------------------------------+----------------------------------+----------------------------+
| * Severe disease cutoffs (NYHA)    | * Anatomical phenotype (DISE)    | * Standardized GDMT wash-in|
| * Quantitative imaging biomarkers  | * ECG morphology (LBBB gating)   | * Stable baseline symptom  |
| * High surgical risk (STS score)   | * Lesion geometry / calcification|   score observation period |
| * Multi-vessel / complex disease   | * Device-specific CDx algorithm  | * Echo core lab calibration|
+------------------------------------+----------------------------------+----------------------------+
                                  │                                  │
                                  ▼                                  ▼
                    +------------------------------------------------------+
                    |              DEVICE-UNIQUE RUN-IN GATES              |
                    | * Temporary percutaneous screening trial (e.g., SCS)|
                    | * Acute hemodynamic response testing (e.g., LVAD)    |
                    +------------------------------------------------------+

1. Prognostic Enrichment: Event-Rate Optimization

Prognostic enrichment selects patients who have a high likelihood of experiencing the clinical event that comprises the primary endpoint, regardless of the treatment assigned. This strategy is critical for event-driven cardiovascular, neurovascular, and oncology device studies where the primary endpoint is a composite of major adverse events (e.g., all-cause mortality, myocardial infarction, stroke, or unplanned rehospitalization).

If a device sponsor designs a trial with an unselected mild-to-moderate disease cohort where the annualized control-arm event rate is only 5%, demonstrating a 25% relative risk reduction (RRR) requires massive patient numbers. By prognostically enriching the trial for high-risk patients—such as requiring a prior hospitalization within 12 months, elevated NT-proBNP levels, or specific echocardiographic thresholds—the control event rate can be increased to 20% or 30%, dramatically shrinking the required sample size and study duration.

2. Predictive Enrichment: Mechanism-Matched Selection

Predictive enrichment identifies and enrolls patients who are more likely to respond favorably to the investigational device compared to patients without the specific selection characteristic. Unlike prognostic markers (which indicate overall disease trajectory), predictive markers reflect a biological, physical, or anatomical mechanism of interaction between the patient and the device.

In pharmaceutical development, predictive enrichment typically relies on genomic, proteomic, or molecular biomarkers (such as HER2 overexpression or PD-L1 expression levels), often requiring formal co-development under FDA's companion diagnostics regulatory framework. In medical devices, predictive enrichment is predominantly driven by:

• Anatomical Phenotyping: Selecting specific vessel dimensions, annular calcification patterns, or airway obstruction morphologies.
• Electrophysiological Gating: Selecting specific conduction abnormalities (e.g., wide QRS with Left Bundle Branch Block) that respond to resynchronization energy.
• Biomechanical Properties: Selecting tissue compliance, bone mineral density thresholds, or joint kinematics compatible with device physics.

3. Variability Reduction: Eliminating Measurement & Clinical Noise

Variability reduction strategies do not seek to change the event rate or isolate super-responders; rather, they aim to decrease measurement error and baseline outcome variance ($\sigma^2$) in continuous endpoints, such as functional exercise capacity (6-Minute Walk Distance [6MWD]), Patient-Reported Outcomes (PROs, such as the Kansas City Cardiomyopathy Questionnaire [KCCQ]), or visual analog pain scales.

Key variability-reduction mechanisms in device trials include:

1. Guideline-Directed Medical Therapy (GDMT) Run-In: Enrolling only patients whose background medical therapy has been optimized and documented as stable for at least 30 to 90 days, preventing spontaneous symptom improvements from confounding device effects.
2. Centralized Core Laboratory Adjudication: Replacing site-level visual interpretations with standardized, blinded core-lab measurements for echocardiograms, CT angiograms, or MRIs.
3. Repeated Baseline Assessments: Averaging multiple baseline functional tests across two separate screening visits to account for regression to the mean and day-to-day patient fluctuations.

---

How Much Does Enrichment Reduce Sample Size? The Event-Rate-to-N Equation

The quantitative power of prognostic enrichment lies in its ability to alter the underlying event rate parameters in sample size formulas. For a two-arm randomized controlled trial comparing a device against a control with a binary or time-to-event primary endpoint, the required sample size per arm ($n$) using the standard two-proportion comparison formula (with two-sided significance level $\alpha = 0.05$ and statistical power $1 - \beta = 0.80$) is given by:

$$n = \frac{\left(z_{1-\alpha/2}\sqrt{2\bar{p}\bar{q}} + z_{1-\beta}\sqrt{p_1 q_1 + p_2 q_2}\right)^2}{(p_1 - p_2)^2}$$

Where:

• $p_1$ is the control-arm event rate.
• $p_2 = p_1 \times (1 - \text{RRR})$ is the device-arm event rate under an assumed Relative Risk Reduction ($\text{RRR}$).
• $\bar{p} = \frac{p_1 + p_2}{2}$ and $\bar{q} = 1 - \bar{p}$.
• $z_{1-\alpha/2} = 1.960$ for a two-sided $\alpha = 0.05$.
• $z_{1-\beta} = 0.842$ for $80%$ statistical power.

Sample Size Reduction Under Prognostic Enrichment (25% RRR, 80% Power, $\alpha=0.05$)

The following table demonstrates the impact of elevating the baseline control-arm event rate ($p_1$) through prognostic enrichment while maintaining a constant 25% treatment effect size ($\text{RRR} = 0.25$):

Baseline Control Event Rate ($p_1$)	Device Event Rate ($p_2$ at 25% RRR)	Absolute Risk Difference ($\Delta$)	Required $N$ per Arm (Pooled Formula)	Total Required Trial Size ($2N$)	Sample Size Multiplier vs. 30% Baseline
5.0%	3.75%	1.25%	4,202	8,404	7.78x
10.0%	7.50%	2.50%	2,005	4,010	3.71x
15.0%	11.25%	3.75%	1,272	2,544	2.36x
20.0%	15.00%	5.00%	906	1,812	1.68x
25.0%	18.75%	6.25%	686	1,372	1.27x
30.0%	22.50%	7.50%	540	1,080	1.00x (Baseline)

Practical Biostatistical Implications

As demonstrated above, moving from an unselected trial population with a 10% control event rate to a prognostically enriched population with a 30% control event rate reduces the required total enrollment from 4,010 patients down to 1,080 patients—a 3.7x reduction in total sample size.

In medical device development, per-patient pivotal-trial costs (including device manufacturing, procedural hospital fees, imaging core-lab reads, and Clinical Research Associate monitoring) are commonly estimated between $50,000 and $120,000 per enrolled patient (MedDeviceGuide planning estimate based on published CRO and industry benchmarks). At that cost range, the 2,930 enrollments avoided above represent roughly $150 million to $350 million in avoided study spend, and can accelerate trial completion by multiple years. Biostatisticians modeling these parameters should cross-reference our comprehensive medical device sample size calculation guide for exact formulas handling attrition, unblinded sample size re-estimation, and survival-analysis models.

---

Recommended Reading

Superiority and Equivalence Clinical Trial Designs for Medical Devices

Clinical Evidence Regulatory2026-08-10 · 27 min read

Which Enrichment Tools Are Unique to Medical Devices?

While pharmaceutical enrichment predominantly centers on molecular assays and genetic biomarkers, medical devices interact with human anatomy, biomechanics, and electrophysiology. Consequently, medical technology has evolved four distinct, device-unique enrichment mechanisms.

+----------------------------------------------------------------------------------------------------+
|                                FOUR DEVICE-UNIQUE ENRICHMENT GATES                                 |
+------------------------------------+----------------------------------+----------------------------+
|   1. TEMPORARY RUN-IN SCREENING    |   2. ENDOSCOPIC & ANATOMICAL     |  3. ELECTROPHYSIOLOGICAL   |
| * 3-7 day percutaneous trial phase | * Drug-Induced Sleep Endoscopy   | * LBBB vs non-LBBB gating  |
| * >=50% pain reduction threshold   | * Complete concentric collapse   | * QRS duration thresholds  |
| * Eliminates acute non-responders  |   exclusion (e.g., Inspire STAR) | * Sex-specific interaction |
+------------------------------------+----------------------------------+----------------------------+
                                     │
                                     ▼
                    +--------------------------------------------------+
                    |        4. QUANTITATIVE CORE-LAB GATING           |
                    | * Echo core lab EROA / regurgitant volume cutoffs|
                    | * Standardized 3D CT annular sizing & calcium    |
                    | * Centralized eligibility adjudication committee |
                    +--------------------------------------------------+

1. Temporary Screening Trials: Spinal Cord Stimulation (SCS)

In implantable neuromodulation for chronic intractable neuropathic pain, permanent device implantation is preceded by a temporary screening trial—a physiological run-in phase with no true pharmaceutical equivalent.

• The Mechanism: Patients undergo percutaneous placement of temporary epidural stimulation leads connected to an external pulse generator for a 3- to 7-day evaluation period in their daily living environment.
• The Enrichment Gate: Only patients who demonstrate objective clinical response—defined as a $\ge 50%$ reduction in visual analog scale (VAS) or numerical rating scale (NRS) pain scores without adverse side effects—are permitted to proceed to permanent surgical implantation and randomization.
• Trial Evidence & Pass Rates: In landmark neuromodulation RCTs (as reviewed by Eldabe et al., Pain, 2020), screening trial pass rates among enrolled patients range between 88% and 93%.
• Regulatory & Coverage Reality: FDA CDRH requires labeling that reflects the trial period, and commercial payers (including CMS National Coverage Determinations) mandate documented success during a temporary trial phase as a prerequisite for permanent implant reimbursement.
• Methodological Caveat: While trial-period enrichment ensures that permanently implanted patients have confirmed responsiveness, it introduces selection bias. Clinical study reports must transparently document all screen failures to prevent overstating device performance in an unselected intention-to-treat population. Notably, the same TRIAL-STIM analysis concluded that a screening-trial strategy provided no superior 6-month pain outcomes versus direct implantation and was not cost-effective in the UK NHS setting — a reminder that device-unique responder gates rest substantially on convention and payer requirements, not solely on demonstrated predictive validity.

2. Anatomical & Endoscopic Phenotyping: Hypoglossal Nerve Stimulation

The pivotal evaluation of the Inspire Upper Airway Stimulation system (the STAR Trial; Strollo et al., N Engl J Med, 2014) represents the gold standard in multi-gate anatomical predictive enrichment.

ALL-COMERS SCREENING POPULATION (Moderate-to-Severe OSA)
  │
  ├── Gate 1: Clinical Failure & Tolerance (CPAP failure/intolerance)
  │
  ├── Gate 2: Biometric & Severity Funnel (BMI ≤ 32 kg/m², baseline AHI 20–50 events/hr)
  │
  ├── Gate 3: Polysomnographic Morphology (Central/mixed apneas < 25% of total AHI)
  │
  └── Gate 4: Drug-Induced Sleep Endoscopy (DISE)
               └── Absence of Complete Concentric Collapse (CCC) at retropalatal level
  │
  ▼
FINAL ENRICHED PIVOTAL COHORT (N = 126; 66% 12-Month Surgical Responder Rate)

1. Gate 1 (Clinical Failure): Documented intolerance or failure of continuous positive airway pressure (CPAP).
2. Gate 2 (Biometric Gating): Body Mass Index ($\text{BMI}$) $\le 32\text{ kg/m}^2$ (patients with higher BMI have increased parapharyngeal fat deposition that overcomes neurostimulation forces).
3. Gate 3 (Severity & Polysomnographic Gating): Baseline Apnea-Hypopnea Index (AHI) between 20 and 50 events per hour, with central and mixed apneas accounting for $<25%$ of the total AHI.
4. Gate 4 (Endoscopic Anatomical Gating): Pre-implantation Drug-Induced Sleep Endoscopy (DISE) to visualize velopharyngeal collapse patterns. Patients exhibiting complete concentric collapse (CCC) of the retropalatal airway were excluded, as hypoglossal nerve stimulation produces anterior tongue displacement that resolves anterior-posterior collapse but fails against circumferential collapse.

The Strategic Outcome: By enforcing this rigorous four-tier anatomical funnel, the STAR trial achieved a 66% primary surgical responder rate at 12 months in $N=126$ patients, securing FDA Premarket Approval (PMA P130008) on April 30, 2014. As detailed later in this guide, the sponsor subsequently expanded this indication to BMI $\le 40\text{ kg/m}^2$ and AHI up to 100 events/hr using real-world registry evidence.

3. Electrophysiological & Morphological Gating: Cardiac Resynchronization Therapy (CRT)

Cardiac Resynchronization Therapy provides a classic demonstration of how post-hoc predictive enrichment findings can evolve into permanent, mandatory guideline criteria.

In the MADIT-CRT trial (Moss et al., N Engl J Med, 2009; Arshad et al., J Am Coll Cardiol, 2011), 1,820 patients with New York Heart Association (NYHA) Class I/II heart failure, $\text{LVEF} \le 30%$, and $\text{QRS duration} \ge 130\text{ ms}$ were randomized to CRT-D or ICD alone. While overall trial results were positive, prespecified and post-hoc predictive analyses revealed massive heterogeneity based on QRS morphology:

• Left Bundle Branch Block (LBBB) Cohort: Patients with true LBBB derived profound clinical benefit from resynchronization, with dramatic reductions in heart failure events and mortality.
• Sex-Morphology Interaction: Women with LBBB demonstrated exceptional responsiveness: in women with LBBB and QRS duration of 130 to 149 ms, CRT-D was associated with a Hazard Ratio of 0.24 ($95%\text{ CI } 0.11–0.53, P<0.001$), representing a 76% reduction in heart failure or death. Across all LBBB patients, women achieved an overall $\text{HR of } 0.31$ (a 69% risk reduction).
• Non-LBBB Cohort: Patients with Right Bundle Branch Block (RBBB) or non-specific intraventricular conduction delay (IVCD) showed no significant benefit and experienced potential harm from biventricular pacing.

Regulatory & Clinical Impact: As a consequence of these predictive findings, international clinical guidelines (ACC/AHA/ESC) and CMS reimbursement policies concentrated Class I CRT indications among patients with LBBB morphology, establishing electrophysiological morphology as a non-negotiable predictive enrichment gate.

---

Why Did COAPT Succeed While MITRA-FR Failed with the Same Device?

The most famous, instructive case study in medical device clinical trial design is the dramatic divergence between the COAPT and MITRA-FR trials. Both studies evaluated the exact same transcatheter edge-to-edge repair (TEER) device (the Abbott MitraClip) in patients with secondary (functional) mitral regurgitation (SMR) and heart failure. Both were published in the New England Journal of Medicine in the exact same December 2018 issue. Yet, one trial was a triumphant success that transformed clinical practice, while the other was completely neutral.

The entire difference came down to patient selection and enrichment criteria.

+----------------------------------------------------------------------------------------------------+
|                         THE SECONDARY MITRAL REGURGITATION DIVERGENCE                              |
+------------------------------------+----------------------------------+----------------------------+
|       MITRA-FR TRIAL (NEJM 2018)   |       COAPT TRIAL (NEJM 2018)    |  GRAYBURN'S DISPROPORTION  |
+------------------------------------+----------------------------------+----------------------------+
| * Looser EU MR criteria:           | * Stricter US MR criteria:       | * MITRA-FR = Proportionate |
|   EROA >= 20 mm² or RV >= 30 mL    |   EROA >= 30 mm² or RV >= 45 mL  |   MR (LV was the primary   |
| * Mean EROA: 31 mm²                | * Mean EROA: 41 mm²              |   disease, clip ineffective|
| * Mean LVEDVi: 135 mL/m²           | * Mean LVEDVi: 101 mL/m²         |                            |
| * Permissive GDMT adjustments      | * Centrally adjudicated, stable  | * COAPT = Disproportionate |
| * 12-Mo Death/HFH: 54.6% vs 51.3%  |   maximally tolerated GDMT       |   MR (MR was disproportion-|
|   (OR 1.16, P = 0.53) - NEUTRAL    | * 2-Yr HFH: 35.8% vs 67.9%/pt-yr |   ate to LV dilation; clip |
|                                    |   (HR 0.53, P < 0.001) - POSITIVE|   altered disease course)  |
+------------------------------------+----------------------------------+----------------------------+

Comprehensive Comparison: COAPT vs. MITRA-FR Protocol & Outcome Parameters

The following table details the protocol selection criteria, baseline patient characteristics, and clinical outcomes across both investigations:

Trial Parameter	MITRA-FR (Obadia et al., NEJM 2018)	COAPT (Stone et al., NEJM 2018)	Clinical & Methodological Impact
Investigational Device	MitraClip Delivery System	MitraClip Delivery System	Identical physical device evaluated in both trials.
Study Geography & Centers	37 centers across France	78 centers across US and Canada	Multi-center randomized controlled trials.
Total Sample Size ($N$)	$N = 304$ (152 Device, 152 Control)	$N = 614$ (302 Device, 312 Control)	COAPT enrolled 2x larger cohort with 2-year follow-up.
Mitral Regurgitation Inclusion Gate	$\text{EROA} > 20\text{ mm}^2$ and/or $\text{Regurgitant Volume} > 30\text{ mL}$ (European Guidelines)	$\text{EROA} \ge 30\text{ mm}^2$ and/or $\text{Regurgitant Volume} \ge 45\text{ mL}$ (US/ASE Guidelines)	Prognostic & Predictive Gate: COAPT required 50% more severe regurgitation at baseline.
Mean Baseline EROA	$31 \pm 10\text{ mm}^2$	$41 \pm 15\text{ mm}^2$	COAPT enrolled patients with 32% larger regurgitant orifice area.
Left Ventricular End-Diastolic Volume Index (LVEDVi)	$135 \pm 35\text{ mL/m}^2$ (severely dilated ventricles)	$101 \pm 34\text{ mL/m}^2$ (moderately dilated ventricles)	MITRA-FR enrolled larger, more end-stage, dilated ventricles.
Background Medical Therapy (GDMT)	Real-world standard care; medication changes permitted during trial	Mandatory, centrally adjudicated maximally tolerated GDMT with documented stability before enrollment	Variability Reduction: COAPT eliminated confounding from post-randomization medical therapy optimization.
Primary Efficacy Endpoint	12-month composite of all-cause mortality or unplanned heart failure hospitalization	2-year annualized rate of all-cause heart failure hospitalizations	Event-driven primary endpoint design.
Primary Outcome Result	54.6% vs. 51.3% Odds Ratio ($\text{OR}$) = 1.16 ($95%\text{ CI } 0.73–1.84$) $P = 0.53$ (Neutral)	35.8% vs. 67.9% per patient-year Hazard Ratio ($\text{HR}$) = 0.53 ($95%\text{ CI } 0.40–0.70$) $P < 0.001$ (Highly Superior)	COAPT demonstrated a 47% relative reduction in heart failure hospitalizations.
All-Cause Mortality	12-month: 24.3% vs. 22.4% ($\text{HR } 1.11, P=0.70$)	24-month: 29.1% vs. 46.1% ($\text{HR } 0.62, P<0.001$)	COAPT demonstrated a 38% relative reduction in mortality.

Grayburn's Conceptual Framework: Proportionate vs. Disproportionate MR

To resolve the apparent contradiction between COAPT and MITRA-FR, cardiovascular researchers led by Paul Grayburn (Circulation, 2019) developed the Proportionate vs. Disproportionate MR Framework:

1. Proportionate MR (The MITRA-FR Cohort): In secondary MR, left ventricular dilation naturally pulls the mitral leaflets apart. In MITRA-FR, the severe LV dilation ($\text{LVEDVi } 135\text{ mL/m}^2$) fully accounted for the moderate degree of MR ($\text{EROA } 31\text{ mm}^2$). The primary disease was end-stage cardiomyopathy; the mitral regurgitation was merely a bystander symptom. Mechanically repairing the valve with a clip could not fix the underlying, failing myocardium.
2. Disproportionate MR (The COAPT Cohort): In COAPT, patients had less dilated ventricles ($\text{LVEDVi } 101\text{ mL/m}^2$) but significantly more severe regurgitation ($\text{EROA } 41\text{ mm}^2$). The regurgitation was disproportionately severe relative to the ventricular size. In these enriched patients, the mitral regurgitation was an active driver of clinical deterioration, and mechanical repair altered the natural history of the disease.
3. The Subgroup Verification: When investigators evaluated the subset of patients in COAPT who had proportionate MR ($\text{EROA/LVEDV}$ ratio $< 0.14$), their clinical outcomes mirrored MITRA-FR—confirming that the patient selection criteria was the treatment effect.

Core Lesson for Device Developers: Setting inclusion cutoffs is not an administrative exercise. An overly broad, unenriched inclusion threshold (as in MITRA-FR) will dilute treatment effects and cause trial failure, while a precisely enriched physiological threshold (as in COAPT) can demonstrate dramatic therapeutic benefit and secure global regulatory approvals.

---

How Did TAVR Turn Enrichment into a Sequential Approval Strategy?

A frequent objection to clinical trial enrichment is that narrowing inclusion criteria limits the initial addressable market. The counter-strategy is the Indication-Expansion Ladder—a multi-stage regulatory lifecycle approach perfected by transcatheter aortic valve replacement (TAVR) manufacturers.

+----------------------------------------------------------------------------------------------------+
|                                    THE TAVR ENRICHMENT LADDER                                      |
+------------------------------------+----------------------------------+----------------------------+
| STAGE 1: EXTREME ENRICHMENT (2010) | STAGE 2: HIGH SURGICAL RISK(2011)| STAGE 3: INTERMEDIATE(2016)|
| * PARTNER 1B (N=358, Inoperable)   | * PARTNER 1A (N=699, High Risk)  | * PARTNER 2A (N=2,032)     |
| * 1-Yr Death: 30.7% vs 50.7%       | * Non-inferior to SAVR           | * STS Score 4% - 8%        |
| * HR 0.55 (P < 0.001)              | * Rapid initial PMA approval     | * Broadened hospital market|
+------------------------------------+----------------------------------+----------------------------+
                                     │
                                     ▼
                    +--------------------------------------------------+
                    | STAGE 4: LOW RISK / NEAR-ALL-COMERS (2019-2026)  |
                    | * PARTNER 3 (N=1,000, Low Risk; mean STS 1.9%)   |
                    | * 1-Yr Death/Stroke/Rehosp: 8.5% vs 15.1%        |
                    | * 7-Year NEJM 2026 Follow-up: 34.6% vs 37.2%     |
                    | * Full commercial market dominance achieved      |
                    +--------------------------------------------------+

The Four Rungs of the TAVR Approval Ladder

1. Rung 1: Extreme Prognostic Enrichment (Inoperable Patients, 2010):
   • Trial: PARTNER Cohort B (Leon et al., N Engl J Med, 2010; $N=358$).
   • Enrichment Gate: Enrolled exclusively patients with severe aortic stenosis who were deemed inoperable by two independent cardiac surgeons (predicted risk of surgical death or severe irreversible morbidity $\ge 50%$).
   • The Payoff: The control arm experienced a massive 50.7% 1-year all-cause mortality rate under standard medical therapy. TAVR reduced 1-year mortality to 30.7% ($\text{HR } 0.55, 95%\text{ CI } 0.40–0.74, P<0.001$). This extreme event rate allowed a definitive, practice-changing pivotal trial with only 358 patients.
2. Rung 2: High Surgical Risk (2011):
   • Trial: PARTNER Cohort A (Smith et al., N Engl J Med, 2011; $N=699$).
   • Enrichment Gate: Patients with predicted surgical mortality $\ge 15%$ (mean Society of Thoracic Surgeons [STS] score $\approx 11.8%$). Demonstrated non-inferiority to surgical aortic valve replacement (SAVR).
3. Rung 3: Intermediate Surgical Risk (2016):
   • Trial: PARTNER 2A (Leon et al., N Engl J Med, 2016; $N=2,032$).
   • Enrichment Gate: Predicted surgical risk between 4% and 8% (mean STS score $\approx 5.8%$). Demonstrated non-inferiority and procedural superiority for transfemoral access.
4. Rung 4: Low Risk / Near All-Comers (2019–2026):
   • Trial: PARTNER 3 (Mack et al., N Engl J Med, 2019; Leon et al., N Engl J Med, 2026; $N=1,000$).
   • Enrichment Gate: Mean STS score $1.9%$. The 1-year primary composite endpoint of death, stroke, or rehospitalization was 8.5% with TAVR vs. 15.1% with surgery ($P<0.001$ for noninferiority, $P=0.001$ for superiority). At 7 years of follow-up (published in NEJM in February 2026), composite outcomes remained comparable (34.6% vs. 37.2%).

Strategic Takeaway for MedTech Pipeline Strategy

Never attempt to conquer an entire addressable market in a single pivotal trial. Deploy extreme prognostic and predictive enrichment to secure your initial PMA approval with minimal sample size, manageable capital expenditure, and high statistical confidence. Once cash flow, physician familiarity, and reimbursement codes are established, use sequential randomized trials or post-market real-world evidence registries to broaden the label across lower-risk cohorts.

---

Recommended Reading

Pure Global AI: A Practical Workflow Guide for Medical Device Regulatory Teams

Regulatory Digital Health & AI2026-08-10 · 21 min read

How Do FDA, FDORA Diversity Mandates, and ICH E20 Treat Enrichment?

Medical device regulatory professionals face a complex, fragmented regulatory landscape when implementing enrichment strategies.

+----------------------------------------------------------------------------------------------------+
|                             REGULATORY LANDSCAPE & ENRICHMENT MATRIX                               |
+------------------------------------+----------------------------------+----------------------------+
|         FDA CDRH FRAMEWORK         |        FDORA SECTION 3601        |     GLOBAL HARMONIZATION   |
|   (Pivotal Design Guidance 2013)   |  (Diversity Action Plans - DAPs) |     (ICH E20 & EU MDR)     |
+------------------------------------+----------------------------------+----------------------------+
| * Demands target-population        | * Pivotal IDE studies covered   | * ICH E20 Step 2b: Covers  |
|   representativeness (FR 2013-     |   once final guidance effective |   adaptive enrichment and  |
|   26690, Docket FDA-2011-D-0567)   | * Draft guidance restored Feb    |   population modification  |
| * Warns of 'subject selection bias'|   2025; statutory deadline passed| * EU MDR Article 62: Focus |
| * No standalone CDRH enrichment doc| * Dec 2025 FDA Guidance: Align   |   on clinical safety, per- |
|   (relies on CDER 2019 Drug Guide) |   clinical gates with diversity  |   formance & generalizability|
+------------------------------------+----------------------------------+----------------------------+

1. The Regulatory Guidance Asymmetry

A critical nuance in medical device regulation is that FDA CDRH has never issued a standalone device enrichment guidance.

• FDA’s primary formal enrichment guidance—Enrichment Strategies for Clinical Trials to Support Demonstration of Effectiveness of Human Drugs and Biological Products (final, March 2019; Federal Register 2019-04815; Docket FDA-2017-D-6159)—was issued exclusively by CDER and CBER. While its methodological taxonomy is universally applied, it is non-binding on device divisions.
• Conversely, CDRH’s foundational guidance—Design Considerations for Pivotal Clinical Investigations for Medical Devices (November 7, 2013; Docket FDA-2011-D-0567)—explicitly requires that study subjects "adequately reflect the target population for the device based on specific enrollment criteria" and cautions against "subject selection bias" that could compromise generalizability.

Device sponsors must navigate this tension by demonstrating that enrichment criteria define a distinct, clinically justifiable intended-use subpopulation rather than artificially cherry-picking trial candidates.

2. The Diversity Action Plan (DAP) Mandate: FDORA Section 3601

The passage of the Food and Drug Omnibus Reform Act of 2022 (FDORA) introduced Section 3601, amending Section 520(g)(9) of the FD&C Act to require mandatory Diversity Action Plans (DAPs) for all pivotal medical device clinical investigations.

• Status & Timeline: FDA issued draft DAP guidance in June 2024. Following administrative challenges, the guidance was temporarily removed from FDA's portal in January 2025 and subsequently restored on February 11, 2025 pursuant to federal court order. Although the statutory June 26, 2025 deadline for final guidance passed without a finalized document, FDORA ties mandatory DAP submission to studies that begin enrollment more than 180 days after final guidance publication — so the formal start date remains pending even as many sponsors now prepare and submit DAPs proactively with their IDE applications.
• The Diversity-Enrichment Tension: DAPs mandate enrollment targets across historically underrepresented demographic groups (age, biological sex, race, ethnicity). Sponsors frequently worry that narrowing inclusion criteria through enrichment will violate DAP commitments.
• The Resolution (FDA December 2025 Guidance): In December 2025, FDA finalized Enhancing Participation in Clinical Trials — Eligibility Criteria, Enrollment Practices, and Trial Designs (CDER/CBER; the guidance applies to drugs and biologics regulated as drugs and superseded the November 2020 Enhancing the Diversity of Clinical Trial Populations guidance). Its section on broadening eligibility in trials using enrichment strategies states that "in general, enrichment should not lead to the exclusion of demographic groups" and that even in enriched trials, "efforts to maintain enrollment criteria that are as broad and representative as possible are important" — and it advises sometimes enrolling a reasonable sample of marker-negative participants to preserve information on the broader disease population. The practical device translation: justify every gate on objective clinical, anatomical, and pathophysiological mechanisms rather than demographic proxies, and design the protocol so the enriched cohort remains as demographically representative as the science allows.

Practical Framework: Reconciling Enrichment Gates with DAPs

Enrichment Gate	Potential Diversity Risk	Compliant Protocol Design Strategy
Anatomical Sizing (e.g., vessel diameter $\ge 6.0\text{ mm}$)	Unintentionally excludes female patients who have smaller average vessel calibers.	Design multi-size delivery catheters or specify vessel-to-device ratio criteria rather than fixed millimeter cutoffs.
Renal Function Thresholds (e.g., eGFR $> 60\text{ mL/min}$)	Disproportionately excludes older adults and African American populations with higher baseline chronic kidney disease.	Adopt tiered eGFR thresholds (e.g., eGFR $\ge 30\text{ mL/min}$) unless contrast-induced nephropathy risk directly precludes use.
Functional Baseline Gates (e.g., 6MWD $> 300\text{ meters}$)	Excludes frail, elderly, or mobility-impaired demographics.	Utilize upper-body ergometry, submaximal exercise protocols, or validated Patient-Reported Outcome measures.
Prior Treatment Failure (e.g., 3 failed medication lines)	Excludes socioeconomically disadvantaged patients with limited healthcare access.	Define failure as documented clinical intolerance or contraindication in addition to formal trial failures.

For a complete breakdown of operational DAP submission workflows, consult our dedicated medical device Diversity Action Plan enrollment guide.

3. International Harmonization: ICH E20 & EU MDR Article 62

For sponsors conducting multinational clinical investigations across the US, European Union, and Asia:

• ICH E20 (Adaptive Designs for Clinical Trials): Endorsed at Step 2b on June 25, 2025, ICH E20 establishes international standards for adaptive clinical trials, including adaptive population enrichment (where patient eligibility is modified mid-trial based on pre-specified interim analyses). Sponsors planning adaptive interim enrichment must align with ICH E20 and establish independent Data Monitoring Committees (DMCs), as detailed in our adaptive clinical trial designs guide.
• EU MDR Article 62 & MDCG Guidelines: Under the European Medical Device Regulation (EU MDR 2017/745), clinical investigations must demonstrate clinical safety, performance, and benefit-risk acceptability. Notified Bodies scrutinize heavily enriched trial data to evaluate whether the clinical evidence is representative of the European patient population. If an enriched trial is used for CE marking, the Summary of Safety and Clinical Performance (SSCP) and Instructions for Use (IFU) must explicitly restrict intended use to the enriched sub-population.

---

How Do You Operationalize Enrichment: Screening Funnels, Core Labs, and SAPs?

Implementing an enrichment strategy introduces significant operational complexity that must be managed from study inception.

+----------------------------------------------------------------------------------------------------+
|                         ENRICHMENT TRIAL OPERATIONAL WORKFLOW                                      |
+------------------------------------+----------------------------------+----------------------------+
|     1. SCREEN-FAILURE LOGISTICS    |     2. CORE-LAB ADJUDICATION     |      3. SAP & SUBGROUPS    |
| * Model 3:1 to 8:1 screening ratios| * Independent, blinded core lab  | * Prospectively define all |
| * Track reasons for screen failure | * Quantitative imaging workflows |   subgroup hypotheses      |
| * Tiered screening budgets         | * Central eligibility clearance  | * Formal alpha splitting   |
+------------------------------------+----------------------------------+----------------------------+
                                     │
                                     ▼
                    +--------------------------------------------------+
                    |        4. FDA PRE-SUBMISSION (Q-SUB) STRATEGY    |
                    | * Formulate targeted questions on inclusion gates|
                    | * Defend clinical necessity of narrow criteria   |
                    | * Propose post-market expansion commitment       |
                    +--------------------------------------------------+

1. Modeling the Screening Funnel and Budget Impact

Heavily enriched trials require screening significantly more patients to identify eligible candidates. While the final enrolled sample size ($N$) is smaller, the screen-to-enrollment ratio increases:

$$\text{Total Patients Screened} = N_{\text{enrolled}} \times \text{Screen-to-Enrollment Ratio}$$

An all-comers device protocol typically loses only a modest fraction of screened candidates, but every additional enrichment gate independently removes patients before randomization. Multi-gate enriched protocols — where severity, anatomical, and core-lab gates are stacked (as in STAR or COAPT) — are commonly planned against screen-to-enrollment ratios of roughly 3:1 to 8:1. Because these ratios are protocol-specific, sponsors should model the funnel gate by gate from early feasibility data rather than assuming an industry average.

Budgeting Rule: Sponsors must establish tiered site compensation models that reimburse investigation sites for screening diagnostic tests (e.g., screening echocardiograms, CT scans, sleep studies) even when the patient fails final eligibility criteria. Failing to compensate sites for screen failures leads directly to site disengagement and enrollment stagnation.

2. Independent Core Laboratories as Enrichment Gatekeepers

To prevent site-level investigator bias and ensure high protocol adherence:

1. Blinded Central Eligibility Verification: In protocols with quantitative imaging thresholds (e.g., aortic annular area, ejection fraction, EROA), clinical sites must upload raw DICOM imaging data to an independent core laboratory. The core lab must issue a formal eligibility certification prior to patient randomization.
2. Standardized Imaging Charters: Establish explicit measurement guidelines, vendor-neutral calibration standards, and inter-rater reliability protocols to minimize measurement variance across participating clinical sites.

3. Pre-specifying Subgroups in the Statistical Analysis Plan (SAP)

If a sponsor anticipates that a device may perform differently across specific physiological strata, the Statistical Analysis Plan (SAP) must prospectively define these hypotheses before trial unblinding:

• Hierarchical Hypothesis Testing: If testing efficacy in both the overall population and an enriched subpopulation, utilize formal gatekeeping or fixed-sequence testing procedures to strictly control trial-wide Type I error ($\alpha = 0.05$).
• Avoid Post-Hoc Data Dredging: Post-hoc subgroup discoveries (such as the LBBB finding in MADIT-CRT) are hypothesis-generating and cannot support primary PMA approval without corroborating data or a dedicated confirmatory investigation. Consult our clinical trial multiplicity guide for exact alpha-allocation strategies.

4. FDA Pre-Submission (Q-Sub) Engagement Strategy

Prior to finalizing a pivotal IDE protocol with aggressive enrichment gates, sponsors should submit a formal Pre-Submission (Q-Sub) to FDA CDRH. Key questions to include in the briefing package:

1. Does CDRH agree that the proposed prognostic enrichment criteria (e.g., prior hospitalization within 12 months, LVEF ≤ 35%, and NT-proBNP ≥ 1,200 pg/mL) define an appropriate high-risk patient population for evaluating the primary composite endpoint?

2. Does CDRH concur with our proposed core-lab adjudication workflow for confirming anatomical eligibility prior to patient randomization?

3. How does CDRH view the alignment between our proposed clinical enrichment criteria and our FDORA Section 3601 Diversity Action Plan enrollment targets?

---

Frequently Asked Questions (FAQ)

What is the primary difference between enrichment, stratification, and subgroup analysis?

Enrichment is a prospective selection strategy where only patients meeting specific criteria are enrolled in the trial; non-qualifying patients are completely excluded. Stratification is a randomization strategy where all eligible patients are enrolled, but randomization is balanced within predefined subgroups (strata) to ensure equal distribution between device and control arms. Subgroup analysis is an analytical technique where treatment effects are evaluated within specific subsets of the enrolled population during or after study completion.

How do enrichment criteria in a pivotal IDE trial restrict the final approved FDA label?

FDA CDRH approves medical devices for the specific patient population evaluated in the pivotal clinical investigation. If a sponsor enriches a study by enrolling only patients with severe disease (e.g., NYHA Class III/IV, or specific anatomical dimensions), the approved Indications for Use (IFU) in the Premarket Approval (PMA) order will be restricted to that exact population. Marketing the device for broader, unstudied patient populations constitutes off-label promotion.

How do you determine whether prognostic or predictive enrichment is appropriate for a novel device?

Evaluate your primary statistical and clinical risk. If pilot data indicate that the device produces a consistent percentage reduction in events across all patients, but overall event rates are low, deploy prognostic enrichment to increase event counts. If pilot data show that the device produces dramatic benefit in a specific anatomical, mechanical, or physiological phenotype but fails in others, deploy predictive enrichment to isolate the responsive phenotype.

How does enrichment interact with FDA Diversity Action Plan (DAP) requirements?

Under FDORA Section 3601 and FDA's December 2025 guidance on clinical trial diversity, enrichment criteria must be justified strictly by objective pathophysiology, anatomy, or clinical pharmacology. As long as criteria do not rely on demographic proxies (such as race, ethnicity, or arbitrary age cutoffs) and the sponsor implements proactive outreach across diverse clinical sites, enrichment is fully compatible with DAP compliance.

Can a sponsor expand a medical device's indication after completing an enriched pivotal trial?

Yes. Device sponsors routinely utilize a two-stage lifecycle approach: securing initial PMA approval in a narrow, highly enriched cohort (where trial size and risk are minimized), and subsequently expanding the labeled indication to broader populations through post-approval randomized trials or prospective real-world evidence (RWE) registries (such as the TAVR PARTNER ladder or the Inspire ADHERE registry).

---

Recommended Reading

Adaptive Designs for Medical Device Clinical Studies: FDA Guidance, Types, and ICH E20

Clinical Evidence Regulatory2026-08-09 · 23 min read

References & Regulatory Citations

1. U.S. Food and Drug Administration (FDA / CDER / CBER). Enrichment Strategies for Clinical Trials to Support Demonstration of Effectiveness of Human Drugs and Biological Products. Guidance for Industry. Final Guidance. March 2019. [Federal Register 2019-04815; 84 FR 9534; Docket FDA-2017-D-6159].
2. U.S. Food and Drug Administration (FDA / CDRH). Design Considerations for Pivotal Clinical Investigations for Medical Devices. Guidance for Industry and FDA Staff. Final Guidance. November 7, 2013. [Federal Register 2013-26690, Docket FDA-2011-D-0567].
3. U.S. Food and Drug Administration (FDA / CDER / CBER). Enhancing Participation in Clinical Trials — Eligibility Criteria, Enrollment Practices, and Trial Designs. Guidance for Industry. Final Guidance (Clinical/Medical Revision 1). December 2025. Supersedes Enhancing the Diversity of Clinical Trial Populations (November 2020).
4. U.S. Food and Drug Administration (FDA). Diversity Action Plans to Improve Enrollment of Participants from Underrepresented Populations in Clinical Studies. Draft Guidance for Industry. June 2024 (Restored February 2025; FDORA Section 3601).
5. International Council for Harmonisation (ICH). ICH Guideline E20: Adaptive Designs for Clinical Trials. Step 2b Draft Guideline. Endorsed June 25, 2025.
6. Stone GW, Lindenfeld J, Abraham WT, et al. Transcatheter Mitral-Valve Repair in Patients with Heart Failure (COAPT Trial). N Engl J Med. 2018;379(24):2307-2318. PMID: 30280640.
7. Obadia J-M, Messika-Zeitoun D, Leurent G, et al. Percutaneous Repair or Medical Treatment for Secondary Mitral Regurgitation (MITRA-FR Trial). N Engl J Med. 2018;379(24):2297-2306. PMID: 30145927.
8. Grayburn PA, Sannino A, Packer M. Proportionate and Disproportionate Functional Mitral Regurgitation: A New Conceptual Framework That Reconciles the Results of the MITRA-FR and COAPT Trials. Circulation. 2019;139(3):353-362. PMID: 30553663.
9. Leon MB, Smith CR, Mack M, et al. Transcatheter Aortic-Valve Implantation for Aortic Stenosis in Patients Who Cannot Undergo Surgery (PARTNER Cohort B). N Engl J Med. 2010;363(17):1597-1607. PMID: 20961243.
10. Mack MJ, Leon MB, Thourani VH, et al. Transcatheter Aortic-Valve Replacement with a Balloon-Expandable Valve in Low-Risk Patients (PARTNER 3). N Engl J Med. 2019;380(18):1695-1705. PMID: 30875300.
11. Leon MB, Mack MJ, Hahn RT, et al. Transcatheter or Surgical Aortic-Valve Replacement in Low-Risk Patients at 7 Years. N Engl J Med. 2026;394(8):773-783.
12. Strollo PJ, Soose RJ, Maurer JT, et al. Upper-Airway Stimulation for Obstructive Sleep Apnea (STAR Trial). N Engl J Med. 2014;370(2):139-149. PMID: 24401053.
13. Eldabe S, Duarte RV, Gulve A, et al. Does a screening trial for spinal cord stimulation in patients with chronic pain of neuropathic origin have clinical utility and cost-effectiveness (TRIAL-STIM)? A randomised controlled trial. Pain. 2020;161(12):2820-2829. PMID: 32618875.
14. Moss AJ, Hall WJ, Cannom DS, et al. Cardiac-Resynchronization Therapy for the Prevention of Heart-Failure Events (MADIT-CRT). N Engl J Med. 2009;361(14):1329-1338. PMID: 19723778.
15. Arshad A, Moss AJ, Foster E, et al. Cardiac resynchronization therapy is more effective in women than in men: the MADIT-CRT trial. J Am Coll Cardiol. 2011;57(7):813-820. PMID: 21310317.