The Time Is Brain Review

Remote ischaemic conditioning in patients with supratentorial intracerebral haemorrhage (RICH-2): a multicentre, randomised, sham-controlled phase 3 trial in China

· DOI: 10.1016/j.eclinm.2026.103900

stroke deep-dive intracerebral-haemorrhage remote-ischaemic-conditioning randomised-trial

Reconstructed on October 14, 2026 for the missed October 11 publication slot. This is a replacement article, not a recovered original. The source was selected in verified PMC look-back mode; it was not recovered from the October 11 archived briefing.

Clinical Question (PICO)

In adults with acute supratentorial intracerebral haemorrhage (ICH) who did not require surgery, does seven days of remote ischaemic conditioning (RIC), added to standard medical management, improve the prespecified proportion with modified Rankin Scale (mRS) 0–2 at 90 days compared with sham conditioning plus standard management?[1]

Bottom Line

In this multicentre, sham-controlled phase 3 trial of 458 participants, RIC did not improve the primary 90-day functional outcome: mRS 0–2 occurred in 156/229 (68.1%) versus 163/229 (71.2%) with sham (adjusted RR 0.97, 95% CI 0.87–1.08; P=0.61).[1] Day-7 haematoma-volume measures favored RIC, but the resolution-rate calculation was post hoc and these imaging findings did not translate into better functional outcomes.[1] Skin petechiae and transient pain were more common with RIC.[1]

Design

  • Trial type: Investigator-initiated, multicentre, randomized, sham-controlled, parallel-group phase 3 trial; participants, treating physicians, outcome assessors, and investigators were blinded.[1]
  • N: 458 randomized 1:1; RIC 229, sham RIC 229.[1] The intention-to-treat (ITT) population included all randomized participants.[1]
  • Setting and enrollment: 20 centers in China; April 22, 2021–April 30, 2024.[1]
  • Randomization: Internet-based, fixed blocks of six, stratified by baseline NIHSS (≤10 vs >10) and haematoma volume (≤20 vs >20 mL).[1]
  • Primary outcome: Prespecified favourable functional outcome, mRS 0–2 at 90 days, analyzed by ITT using robust Poisson regression adjusted for age, baseline NIHSS, haematoma volume, and location.[1] Five missing 90-day mRS outcomes were imputed using the last known value when available; otherwise the assigned extreme was mRS 6 for RIC and 0 for sham.[1]

Population

Inclusion Criteria

  • Adults aged 18–80 with imaging-confirmed spontaneous supratentorial ICH, randomizable and treatable 24–48 hours after symptom onset; NIHSS 6–20, local ABC/2-estimated haematoma volume 10–30 mL, and Glasgow Coma Scale (GCS) score >8.[1]

Exclusion Criteria

  • Surgery performed or planned (including evacuation or decompressive craniectomy); concomitant subarachnoid or intraventricular haemorrhage; an upper-limb condition preventing cuff RIC (e.g. severe soft-tissue injury, fracture, or peripheral vascular disease); or estimated life expectancy under six months from severe comorbidity.[1]

Baseline Characteristics

  • Median age 58 years (IQR 51–68); 134/458 (29.3%) were women.[1] Median NIHSS was 9 (IQR 7–13), median GCS 14 (IQR 13–15), and centrally summarized baseline haematoma volume 12.4 mL (IQR 8.3–17.8); 419/458 (91.5%) had basal-ganglia haemorrhage.[1] All participants were of Han Chinese ethnicity.[1]
  • Median onset-to-randomization was 35.7 hours in RIC and 34.4 hours in sham; onset-to-treatment medians were 35.8 and 34.4 hours, respectively.[1]

Interventions

Both groups received standard medical management.[1] RIC used a unilateral upper-arm cuff at 200 mmHg; sham used 30 mmHg.[1] Each session consisted of five cycles of five minutes inflated and five minutes deflated (45 minutes total), first within 10 minutes of randomization and then once daily for seven consecutive days, administered by trained nurses.[1]

Outcomes

The 90-day mRS 0–2 endpoint was the prespecified primary patient-centered outcome.[1] Adjusted efficacy models used age, baseline NIHSS, haematoma volume, and location.[1] Imaging data were available for 203 RIC and 208 sham participants; they are surrogate/mechanistic outcomes, not substitutes for functional benefit.[1]

Outcome RIC Sham Reported effect
Primary: mRS 0–2 at 90 days (ITT) 156/229 (68.1%) 163/229 (71.2%) Unadjusted RR 0.96 (0.85–1.08), P=0.48; adjusted RR 0.97 (0.87–1.08), P=0.61.
mRS 0–1 at 90 days 124/229 (54.1%) 121/229 (52.8%) Adjusted RR 1.03 (0.89–1.20), P=0.66.
mRS 0–3 at 90 days 188/229 (82.1%) 190/229 (83.0%) Adjusted RR 0.99 (0.92–1.07), P=0.80.
mRS 0–2 at 180 days 169/229 (73.8%) 175/229 (76.4%) Adjusted RR 0.97 (0.88–1.07), P=0.57.
Full mRS distribution at 90 days Median 1 (IQR 1–3) Median 1 (IQR 1–3) Adjusted common OR 0.99 (0.71–1.40), P=0.94.
Full mRS distribution at 180 days Median 1 (IQR 0–3) Median 1 (IQR 0–2) Adjusted common OR 1.02 (0.73–1.45), P=0.90.
mRS 0–1 at 180 days 144/229 (62.9%) 148/229 (64.6%) Adjusted RR 0.98 (0.86–1.11), P=0.70.
mRS 0–3 at 180 days 198/229 (86.5%) 197/229 (86.0%) Adjusted RR 1.00 (0.94–1.08), P=0.90.
NIHSS change from baseline at 24 h / 72 h / 7 d — — Adjusted mean differences −0.12 (−0.50 to 0.26), P=0.53; −0.20 (−0.75 to 0.35), P=0.48; and −0.01 (−0.69 to 0.66), P=0.97, respectively.
Imaging: follow-up haematoma volume, median (IQR) 5.6 mL (3.4–8.2) 6.7 mL (4.0–9.6) Adjusted mean difference −1.80 mL (−2.42 to −1.17), P<0.001.
Imaging: haematoma-volume reduction, median (IQR) 6.9 mL (4.7–9.9) 4.4 mL (3.1–6.8) Adjusted mean difference 2.04 mL (1.45–2.62), P<0.001.
Haematoma resolution rate 58.1% (47.4–66.6) 40.9% (33.7–50.5) Adjusted mean difference 14.07% (11.21–16.94), P<0.001; rate was calculated post hoc from prespecified volume measurements.
Subsequent surgery (exploratory outcome) 2/229 (0.9%) 9/229 (3.9%) Adjusted risk difference −3.3 percentage points (−6.1 to −0.5), P=0.02; few events.
All-cause mortality by 90 days 7/227 (3.1%) 6/226 (2.7%) Adjusted risk difference 0.3 points (−2.7 to 3.3), P=0.84.
All-cause mortality by 180 days 7/225 (3.1%) 9/224 (4.0%) Adjusted risk difference −1.0 points (−4.4 to 2.4), P=0.57.
Haematoma expansion 7/229 (3.1%) 12/229 (5.2%) Adjusted RD −0.15 points (−5.0 to 2.0), P=0.41; transcribed as printed; an inconsistent unadjusted interval is flagged under Criticisms.
Neurologic deterioration at 24 / 48 / 72 h / 7 d 5/229 (2.2%) / 6/229 (2.6%) / 2/229 (0.9%) / 3/229 (1.3%) 2/229 (0.9%) / 4/229 (1.7%) / 2/229 (0.9%) / 1/229 (0.4%) Adjusted RD 1.4 (−0.9 to 3.7), 0.9 (−1.8 to 3.6), 0.0 (−1.7 to 1.8), and 1.1 (−0.6 to 2.8) points; P=0.23, 0.53, 0.97, and 0.21, respectively.
Serious adverse events by 90 days 16/229 (7.0%) 12/229 (5.2%) Adjusted risk difference 1.7 points (−2.7 to 6.0), P=0.45.
Serious adverse events through 180 days 19/229 (8.3%) 20/229 (8.7%) Adjusted risk difference −0.3 points (−5.4 to 4.8), P=0.91.
RIC-related skin petechiae 38/229 (16.6%) 0/229 Adjusted risk difference 16.8 points (12.0–21.6), P<0.001.
RIC-related transient pain 12/229 (5.2%) 1/229 (0.4%) Adjusted risk difference 4.8 points; Table 3 CI 1.9–7.7, P=0.001.

The authors reported no deaths attributed to RIC.[1] For the primary endpoint, five 90-day mRS values were missing and imputed as described above; the 180-day analysis had four missing RIC and five missing sham outcomes.[1]

Figures

RICH-2 participant flow from screening through randomization and follow-up populations.
Figure 1. Trial profile distinguishing the randomized ITT groups from smaller per-protocol follow-up sets.[1]

Source: Zhao et al., eClinicalMedicine (2026), Fig. 1, PMCID 13092074. Reproduced unmodified under the article's CC BY-NC 4.0 license; source article.

Modified Rankin Scale outcome distributions for RIC and sham at 90 and 180 days.
Figure 2. Functional-outcome distributions at 90 and 180 days; the primary comparison did not show a significant benefit.[1]

Source: Zhao et al., eClinicalMedicine (2026), Fig. 2, PMCID 13092074. Reproduced unmodified under the article's CC BY-NC 4.0 license; source article.

Prespecified intention-to-treat subgroup forest plot for the primary functional outcome.
Figure 3. Prespecified subgroup analysis of the primary outcome in the ITT population. It is exploratory for effect modification and does not supersede the overall randomized comparison.[1]

Source: Zhao et al., eClinicalMedicine (2026), Fig. 3, PMCID 13092074. Reproduced unmodified under the article's CC BY-NC 4.0 license; source article.

Criticisms

  • The primary functional result is neutral, not proof of equivalence. Its adjusted RR CI (0.87–1.08) includes modest benefit and harm.[1] The trial planned around 30% favourable outcome in sham and 43% in RIC, but observed about 70% in each arm; the authors note this reduced the effect size anticipated and may have increased type II error.[1]
  • Generalizability is limited: all participants were Han Chinese, fewer than one-third were women, and patients with minor or severe neurologic impairment, IVH, or planned surgery were excluded.[1] Baseline eligibility volume used local ABC/2 estimates, which the authors note may overestimate irregular haematomas.[1]
  • The 30-mmHg sham may not have matched sensations from the 200-mmHg intervention; the authors acknowledge possible partial unblinding and performance bias.[1]
  • Only 411/458 were included in imaging analyses; exclusions included missing follow-up images, haematoma expansion, and surgery.[1] The resolution-rate endpoint was a post-hoc calculation; secondary outcomes were not multiplicity-adjusted, and subsequent surgery was exploratory.[1]
  • Table 3 reporting discrepancy: the unadjusted haematoma-expansion RD is printed as −2.2 percentage points with a 95% CI of −1.5 to 5.8, so the point estimate lies outside its reported interval.[1] The adjusted row is printed as −0.15 points (95% CI −5.0 to 2.0; P=0.41); its asymmetry alone does not establish an error, and no corrected value is inferred.[1] For transient pain, the table gives an adjusted CI of 1.9–7.7 points, while the Discussion reports 1.9–7.6.[1] Values above are transcribed rather than silently repaired.[1]
  • The sample-size section says the maximum target was 452, whereas Results report 458 randomized; the article text reviewed here does not explain the difference.[1]

Funding

Supported by the National Natural Science Foundation of China (grants 82422024 and 82001257), Beijing Natural Science Fund for Outstanding Young Scholars (JQ22020), National Key R&D Program of China (2022YFC2408800), Beijing Nova Program (Z201100006820143), and Capital Medical University Outstanding Young Talents Program (1220090115).[1] The authors state that funders had no role in study design, data collection, analysis, interpretation, or report writing.[1]

The paper

  • Authors. Wenbo Zhao, Jing Wang, Xuebin Yu, Guang Wu, Song Yang, Da Lu, Xinjing Gao, Yuping He, Lijian Niu, Haihang Zhou, Caixia Xuan, Na Li, Mengfei Zhong, Jiangtao Zhang, Qinglin Dong, Xiaokun Geng, Qingfeng Ma, Haiqing Song, Yunjian Yin, Lan Liu, David C. Hess, Craig S. Anderson, Sijie Li, Xunming Ji, and the RICH-2 Investigators.[1]
  • Title. Remote ischaemic conditioning in patients with supratentorial intracerebral haemorrhage (RICH-2): a multicentre, randomised, sham-controlled phase 3 trial in China.[1]
  • Journal. eClinicalMedicine. 2026;95:103900.[1]
  • DOI. 10.1016/j.eclinm.2026.103900.
  • PMCID / PMID. PMC13092074 / 42011250.[1]
  • Registration. ClinicalTrials.gov NCT04657133.[1]
  • License. Creative Commons Attribution-NonCommercial 4.0 International (CC BY-NC 4.0), as stated in the article.[1]
  • Primary source. [1] — Europe PMC full-text JATS.
Deep Dive — click to expand

What this is

RICH-2 randomized a selected, non-surgical supratentorial ICH population to seven days of RIC or sham, with a prespecified functional endpoint at 90 days.[1] It found a radiographic clearance signal without an improvement in the functional outcome the trial was designed to test.[1]

1. Shadow Audit

The mechanistic story and the clinical result diverge: follow-up haematoma volume and reduction favored RIC, yet the primary mRS 0–2 result was 68.1% versus 71.2% (adjusted RR 0.97, 95% CI 0.87–1.08; P=0.61).[1] The haematoma resolution rate was calculated after the fact from prespecified volume measurements.[1] That distinction matters: an imaging change may support biological activity, but it is not evidence that patients function better.[1] The 2-versus-9 subsequent-surgery difference is also a sparse exploratory finding, not a confirmed clinical benefit.[1]

2. Inversion Engine

A claim that RIC improves function in this population would require a reproducible benefit on a prespecified patient-centered endpoint, not only faster clearance or a favorable exploratory event count.[1] The primary CI includes no effect and a range of modest relative benefit or harm; it does not establish equivalence, but it does not support a clinically meaningful benefit as observed.[1] A future study could change confidence by enrolling a population with a different severity profile or testing a different protocol, while retaining blinded outcome assessment and adequate power.[1]

3. Second-Order Catalyst

The immediate consequence is not to adopt RIC as routine ICH therapy.[1] The result instead defines a research question: whether a protocol started earlier, given more frequently, or tested in a population with larger haematomas or planned surgery can convert the imaging signal into functional gain.[1] Those are hypotheses raised by the authors, not effects demonstrated in RICH-2.[1] A decisive next trial should prespecify the population, protocol, functional endpoint, harms, and treatment-effect analysis.[1]

4. Asymmetric Leverage

RIC changed day-7 imaging measures, but those surrogate outcomes and the prespecified disability endpoint answer different questions.[1] The surgery result (2/229 vs 9/229; adjusted RD −3.3 points, P=0.02) could be clinically important if replicated, yet only 11 events occurred and surgery was exploratory among multiple secondary outcomes.[1] The low-cost-looking cuff procedure is not automatically low-burden: petechiae affected 16.6% and transient pain 5.2% of RIC recipients.[1]

5. Paradigm Destroyer

The paper does not invalidate haematoma clearance as a biological target; it challenges treating target engagement as a proxy for patient benefit.[1] A measured protocol lesson is to keep radiographic and functional outcomes separate in trial interpretation.[1] It does not justify a bedside change or an individual treatment recommendation.[1]

MVP — Minimum Viable Proof

The minimum proof for clinical adoption would be a sufficiently powered, well-blinded randomized study showing improvement in a prespecified patient-centered functional outcome in a clearly defined ICH population, with transparent harms.[1] RICH-2 did not meet that efficacy test for its studied regimen and population.[1]

Best Combination

Read the day-7 imaging results alongside—not instead of—the 90- and 180-day mRS outcomes, and keep the prespecified primary analysis distinct from post-hoc resolution rate and exploratory surgery.[1] This paired interpretation recognizes target engagement without promoting an unproven clinical effect.[1]

Overvalue Warning

Do not describe greater haematoma clearance as improved recovery, or treat a nominally significant exploratory surgery result as confirmation of benefit.[1] Conversely, do not generalize this neutral trial to every ICH severity group or every RIC schedule: the trial excluded minor and severe cases, enrolled no non-Han population, and tested one unilateral seven-day protocol.[1]

Educational use notice. The Time Is Brain Review is dedicated to medical education. Figures from published articles are embedded for teaching purposes with attribution to the original publication and under the source article's stated license. We do not claim copyright over embedded figures; rights remain with the original publisher and authors. See our editorial notice for the full policy.