'If It's COVID, Paxlovid? Studies Suggest a Rethink'

 

• Two multicenter trials found no change in hospitalization and death rates when antiviral nirmatrelvir-ritonavir (Paxlovid) was given to COVID-19 patients already mostly vaccinated.
• However, nirmatrelvir-ritonavir was linked to more early sustained recovery when taken within the first 5 days of illness in both trials.
• Additionally, there was evidence of more nirmatrelvir-ritonavir recipients reaching a viral load below detection level at day 5 following treatment.

Oral nirmatrelvir-ritonavir (Paxlovid) failed to spare COVID-19-vaccinated individuals from the worst outcomes if they got sick, though it may have helped speed recovery time and cut viral loads, according to two community-based clinical trials.

In the U.K.-based PANORAMIC trial, the incidence of all-cause hospitalization or death after 28 days was 0.8% among those receiving the antiviral combination plus usual care, which was comparable to the 0.7% of peers receiving only usual care (adjusted OR 1.18, 95% Bayesian credible interval [CrI] 0.55-2.62).

With an estimated 33% chance of superiority over usual care, nirmatrelvir-ritonavir fell well short of the superiority threshold of 97.5% in this trial of over 3,000 higher-risk people.

In the smaller Canadian CanTreatCOVID trial, the same outcome of hospitalization or death reached 0.6% in the nirmatrelvir-ritonavir group and 1.2% with usual care. The between-arm difference was again not significant (adjusted OR 0.48, 95% CrI 0.08-2.23), and the antiviral treatment's 83% probability of superiority over usual care once more fell short of the 97.5% threshold.

The two studies were reported together by Christopher Butler, MD, of the University of Oxford, England, and colleagues in the New England Journal of Medicine.

"We found no evidence that early treatment with nirmatrelvir-ritonavir reduced the already-low incidence of hospitalization or death in either trial and were unable to identify any prespecified subgroup with compelling evidence of treatment effect," Butler and colleagues wrote.

Nirmatrelvir-ritonavir was first made available under emergency use authorization by the FDA in late 2021, and was FDA approved in 2023 as a 5-day regimen for treating adult outpatients with mild to moderate COVID-19 at risk for severe disease. Approval was based on efficacy data in symptomatic, unvaccinated adults testing positive for SARS-CoV-2 infection in the EPIC-HR trial.

Now, the PANORAMIC and CanTreatCOVID results reflect a COVID-19 landscape that's shifted since the pandemic's early period, said H. Clifford Lane, MD, former deputy director for clinical research and special projects at the National Institute of Allergy and Infectious Diseases (NIAID), and Anthony Fauci, MD, the former NIAID director.

"These new data indicate that the 89% relative risk reduction seen in the analysis of hospitalizations or death associated with the use of nirmatrelvir-ritonavir in the EPIC-HR trial does not apply to the current circumstances, in which most adults have varying degrees of preexisting immunity and the circulating variants are different," Lane and Fauci wrote in an accompanying editorial.

That doesn't mean nirmatrelvir-ritonavir's therapeutic time has come and gone, they cautioned. PANORAMIC and CanTreatCOVID participants who took the combination drug saw enhanced recovery and faster viral load reductions, they noted, which points to both clinical efficacy and antiviral activity.

Indeed, Butler's group reported that early sustained recovery, defined as recovery by day 14 of treatment that was sustained to day 28, reached 33% and 22.1% in PANORAMIC's treatment and control arms, respectively (adjusted OR 1.74, 95% CrI 1.48-2.04), and 69% and 53.1% in CanTreatCOVID (adjusted OR 1.99, 95% CrI 1.40-2.87). Median time to recovery with the antiviral was 3 to 7 days shorter in the two studies.

Furthermore, in a PANORAMIC virology substudy of 485 participants, 29.2% of nirmatrelvir-ritonavir recipients reached a viral load below detection level at day 5, a significant improvement over the 16.5% of controls (adjusted OR 2.15, 95% CrI 1.37-3.44). Day 5 geometric mean viral load was 3,587 in the nirmatrelvir-ritonavir group and 30,267 among controls (adjusted geometric mean ratio 0.13, 95% CrI 0.08-0.21).

"Clinicians may become more selective regarding which patients to refer for treatment," Lane and Fauci wrote. "But it still would seem prudent to consider antivirals on a case-by-case basis, particularly in older adults, persons with a compromised immune system, and persons for whom more-rapid recovery is a priority."

The open-label, prospective PANORAMIC and CanTreatCOVID trials enrolled adults in the community at higher risk of COVID-19, which included either those age 50 years or older and younger adults with relevant coexisting conditions. Participants had to have SARS-CoV-2 infection symptoms for 5 days or less and a positive COVID test.

Patients were randomized to receive either usual care plus daily oral nirmatrelvir-ritonavir for 5 days, or usual care alone. The PANORAMIC primary analysis included 1,698 people who received nirmatrelvir-ritonavir and 1,673 who received usual care. CanTreatCOVID included 343 people in the nirmatrelvir-ritonavir group and 324 in the usual-care group.

PANORAMIC evaluated nirmatrelvir-ritonavir from April 2022 to March 2024, while CanTreatCOVID evaluated the intervention from January 2023 to September 2024. Nearly everyone in PANORAMIC and CanTreatCOVID was vaccinated, with rates reaching at least 97.8% among all participants.

Most of PANORAMIC's nirmatrelvir-ritonavir group had adverse events (90.4%), while 0.5% had serious adverse events. The rate of serious adverse events in CanTreatCOVID was higher in the usual-care only group (3.4%) than in the nirmatrelvir–ritonavir group (1.3%).

Butler's group acknowledged that the two studies had an open-label design that precluded estimation of placebo or nocebo effects.

Additionally, the CanTreatCOVID trial was stopped early due to slow recruitment and because the supply of nirmatrelvir-ritonavir was discontinued.

Disclosures

The two studies were supported by the U.K. National Institute for Health and Care Research, the Canadian Institutes of Health Research, and Health Canada.

Butler, Lane, and Fauci had no relevant disclosures. Study coauthors disclosed relationships with multiple companies.

Primary Source

New England Journal of Medicine

Source Reference: Butler CC, et al "Oral nirmatrelvir-ritonavir for Covid-19 in higher-risk outpatients" N Engl J Med 2026; DOI: 10.1056/NEJMoa2502457.

Secondary Source

New England Journal of Medicine

Source Reference: Lane HC, Fauci AS "Same pill, different impact -- reassessing the efficacy of nirmatrelvir-ritonavir" N Engl J Med 2026; DOI: 10.1056/NEJMe2603127.

https://www.medpagetoday.com/infectiousdisease/covid19/120915

'Machine Analyzes Joint X-Rays About as Well as Humans'

 

• The Sharp/van der Heijde (SvdH) method for analyzing X-ray images is the standard way to measure joint space narrowing and bone erosions in rheumatoid arthritis, requiring well-trained readers.
• Recent advances in machine learning and artificial intelligence carry the potential to automate SvdH scoring.
• In this early study, a machine-learning system called autoscoRA produced SvdH scores with good to excellent concordance with experienced human readers.

A machine-learning system for analyzing rheumatoid arthritis (RA) patients' X-rays was able to produce Sharp/van der Heijde (SvdH) scores, the standard way to quantify joint space narrowing and bone erosions, with good accuracy when compared with human readers, researchers said.

Called autoscoRA, the system matched the human reader's scores for joint space narrowing in more than 95% of hand and foot images, according to Thomas Deimel, MD, of the Medical University of Vienna in Austria, and colleagues.

Performance in scoring erosions was more variable, the group reported in Arthritis & Rheumatology, but the level of agreement was still considered good. Scoring differences greater than 1 point in the SvdH method were seen for only 6.3% of hand images and 11.0% of foot X-rays.

Another finding in favor of autoscoRA came from a test in which images were scored by the first human reader, autoscoRA, and a second human reader. AutoscoRA agreed with the first reader for summed scores with an intraclass correlation of 0.94, whereas the second human reader's scores agreed with the first's with a correlation of 0.86. When scoring individual joints, autoscoRA readouts for joint space narrowing differed from the first reader's by more than 1 point in fewer than 3% of instances, whereas for the second reader, about 10% of images had these differences.

"For the erosion score, the performance of the automated system roughly matched that of the second human reader numerically, although visual inspection indicated potentially more consistent predictions by the former," Deimel and colleagues added.

One of the problems with standard SvdH scoring is that inter-reader (and even intra-reader) reliability is only so-so. Consistency, therefore, is a desirable goal for any method of radiograph analysis. For one thing, errors are more likely to be systematic and therefore more easily recognized and rectified than if they occur randomly.

Another reason to favor an automated system is cost and efficiency. SvdH scoring requires considerable training and experienced readers are therefore scarce, especially outside major referral centers. The reading itself is also time-consuming and, with the requirement for specialized staff to do it, expensive. "An automated system such as autoscoRA directly addresses the feasibility gap, offering a scalable and reproducible solution that transforms imaging into reliable, structured outcome data," Deimel and colleagues wrote.

AutoscoRA has been under development for some time; Deimel gave a preliminary presentation on it at a 2020 rheumatology conference. This new study included many more images and additional analyses to better define the system's potential.

The researchers drew on a large archive of hand and foot X-rays from 769 RA patients seen at the Medical University of Vienna, who had a total of 3,437 clinic visits and more than 12,000 radiographs. Some 60% of images were used for training, 20% for validation, and 20% as a "test set." The comparative testing with autoscoRA and human readers was performed on this latter set.

Besides scoring individual radiographs, the study also looked at serial images from 54 patients with a total of 237 visits over time. This allowed an examination of how autoscoRA could quantify disease progression. Agreement with the human reader averaged 70% over a range of progression definitions (i.e., the degree of change in erosion and joint space scores over time). "Overall performance appeared to be relatively stable across the range of cutoffs," the researchers wrote.

Deimel and colleagues stressed that autoscoRA needs additional "external validation" with images from other institutions, as well as more focus on the system's ability to assess progression over time, before it could be considered for routine clinical use. In the meantime, however, the researchers suggested that it could find near-term application in clinical trials and for analyzing large image collections such as those in registries and other observational patient cohorts.

Disclosures

The study was funded by the Innovative Health Initiative Joint Undertaking, which is supported by the European Union, Swiss and Austrian national governments, and a variety of nonprofit organizations and commercial enterprises involved in healthcare technology.

Co-authors reported relationships with numerous pharmaceutical companies.

Primary Source

Arthritis & Rheumatology

Source Reference: Deimel T, et al "autoscoRA: deep learning to automate Sharp/van der Heijde scoring of radiographic damage in rheumatoid arthritis" Arthritis Rheumatol 2026; DOI: 10.1002/art.70196.

https://www.medpagetoday.com/radiology/diagnosticradiology/120923

IBM beats Q1 2026 estimates with EPS $1.91, revenue $15.9B as J.P. Morgan cuts price target

 

IBM beats Q1 2026 estimates with EPS $1.91, revenue $15.9B as J.P. Morgan cuts price target to $270 from $283

• J.P. Morgan Securities cut IBM price target to $270 from $283 after it maintained full-year guidance despite slowing revenue growth.
• Q1 2026 revenue was $15.92B, up 9% YoY, supported by AI-driven hybrid cloud and mainframe growth.
• Q1 revenue grew 6% at constant currency, with broad-based strength in software and infrastructure.
• Software revenue rose 8%, driven by 16% growth in Data and 10% growth in Red Hat.
• Infrastructure revenue increased 12%, including 48% growth in IBM Z and double-digit distributed infrastructure.
• Consulting grew 1%, with signings up 6% and generative AI now ~30% of backlog.
• Consulting revenue slightly missed estimates despite overall Q1 earnings and revenue outperformance.
• Margins expanded strongly: operating pretax margin +140 bps, infrastructure margin +720 bps, software +60 bps.
• Free cash flow was $2.2 billion, up 13% YoY, the strongest Q1 in a decade.
• Full-year 2026 guidance maintained: 5%+ constant-currency revenue growth and ~$1 billion free cash flow increase.
• Management now expects software to grow 10%+ in 2026, data segment low-20%+, consulting low-to-mid single digits.
• IBM increased its quarterly dividend to $1.69, raising its regular payout to shareholders.
• Key risks: macro/geopolitics, infrastructure and mainframe cycle dependence, despite unchanged guidance and no reported demand deterioration.
• Strong quarter, driven by accelerating software and infrastructure growth, expanding margins, and robust free cash flow.

https://finviz.com/quote.ashx?t=IBM&p=d

Prediction Markets Kalshi traders don't see Strait of Hormuz traffic returning to normal until July

 Bettors on the prediction markets platform Kalshi don’t think the Strait of Hormuz will be open to normal traffic flows for months. 

Odds that traffic will return to normal by June 1 fell below 50% on Wednesday, after the U.S. and Iran extended a ceasefire but neither side disclosed any new agreement regarding Iran opening the Strait of Hormuz or the U.S. ending its naval blockade of the passageway. 

On Thursday, President Donald Trump threatened to “shoot and kill” any boat laying mines in the strait, while oil prices climbed higher with Brent crude again above $100 per barrel. 

Bettors on Kalshi give just a 42% chance that normal traffic flows through the strait by June 1. They assign a 59% chance that happens by July 1, and a 61% chance by Aug. 1. Kalshi defines normal traffic flows on the contract as the seven-day moving average of transit calls through the strait based on data from IMF PortWatch. 

On Polymarket, bettors give a 45% chance that traffic through the strait returns to normal by the end of May, and a 67% chance by the end of June. Polymarket uses the same definition of normal traffic as Kalshi. 

Transit through the strait remains low. On Wednesday — the same day Iran said it seized two ships sailing through the strait without authorization — eight ships crossed the strait, including three oil tankers, according to data from LSEG. Before the war, traffic typically included more than 100 ships daily in the strait. 

https://www.cnbc.com/2026/04/23/kalshi-bettors-see-strait-of-hormuz-traffic-normal-by-july.html

WH accuses China of stealing AI technology

 White House Office of Science and Technology Policy (OSTP) Director Michael Kratsios accused China of carrying out "industrial-scale" theft of the United States' artificial intelligence labs' technology in a memo published on Thursday.

"The US government has information indicating that foreign entities, principally based in China, are engaged in deliberate, industrial-scale campaigns to distil US frontier AI systems," Kratsios said in the memo, adding that the campaigns were "leveraging tens of thousands of proxy accounts to evade detection and using jailbreaking techniques to expose proprietary information."

The White House official also revealed that the US government will inform American AI companies about "attempts by foreign actors to conduct unauthorized, industrial-scale distillation" to help them protect themselves from the attacks, and that Washington will consider measures to hold those actors accountable.

https://breakingthenews.net/Article/WH-accuses-China-of-stealing-AI-technology/66136553

US deploys drones in Hormuz mine clearance

 The U.S. military is accelerating mine-clearing operations in the Strait of Hormuz. Maritime drones are reportedly being used in this operation. While specific details remain undisclosed, the U.S. Navy possesses diverse capabilities, including helicopters, littoral combat ships, LCS, and even trained dolphins.

According to the Wall Street Journal, WSJ, on the 19th, a U.S. Department of Defense official stated that the U.S. military is employing both manned and unmanned capabilities in mine-clearing operations.

Maritime drones encompass unmanned surface vessels and unmanned underwater vehicles, capable of detecting mines beneath the sea using underwater sonar without exposing crew members to danger.

As the U.S. Navy retires traditional minesweepers, vessels designed to detect and remove mines, maritime drones are increasingly becoming crucial to the Navy’s anti-mine capabilities, according to the WSJ. The conventional method had the drawback of exposing sailors to mine threats. Scott Savitz, a researcher at the RAND Corporation, a think tank, stated, "Since there is less concern about casualties, sending drones into minefields is much more acceptable," adding, "Even if some are lost, they can be replaced."

The anti-mine capabilities currently operable by the U.S. Navy are diverse. Unmanned surface vessels developed by defense contractor RTX are equipped with the AQS-20, a new floating underwater sonar system, and can sweep a 100-foot (approximately 30-meter) wide seabed at once. Underwater drones developed by General Dynamics, the 'MK18 Mod 2 Kingfish' and 'Knifefish,' use batteries as their power source and can be deployed from small boats.

Notably, the Navy is also known to utilize trained dolphins. According to the U.S. military, the Navy employs bottlenose dolphins and California sea lions in missions such as mine detection through its Marine Mammal Program. These animals are trained to detect, identify, mark, and recover objects in ports, coastal waters, and deep seas. Both species are known for their high trainability and adaptability to various marine environments. Particularly, dolphins can easily locate potential hazards like mines even in environments difficult to detect with sonar, thanks to their exceptional echolocation capabilities. However, it remains unknown whether dolphins were deployed in the current mine-clearing operation in the Strait of Hormuz.

Retired Rear Admiral Kevin Donegan, former commander of the U.S. Navy’s 5th Fleet, projected, "Unmanned underwater vehicles can survey small waterways in the region within days rather than weeks." He added that if the Navy first clears mines in narrow areas to resume partial passage, safe routes can gradually be expanded.

It remains uncertain how many mines Iran has deployed. However, experts analyze that due to U.S. military pressure, Iran likely failed to lay large-scale mines. Bryan Clark, a senior fellow at the Hudson Institute and former senior U.S. Navy official, stated, "Due to U.S. military pressure, Iran probably could not use large mine-laying vessels and had to rely on small fishing boats or cargo ships," adding, "The number of mines is likely smaller than expected."

Even after agreeing to a ceasefire with the U.S., Iran has not fully opened the Strait of Hormuz. Iran has warned that mines are present in major shipping routes of the strait and has demanded the use of new routes under its control. There are observations that if the U.S. clears mines to resume partial passage, it could weaken Iran’s control over the strait and serve as a driving force for Iran to engage more actively in negotiations.

https://www.msn.com/en-xl/news/other/us-deploys-drones-in-hormuz-mine-clearance/ar-AA21ifW1

UK Biobank patient data stolen and placed on sale in China

 There has been a major data security lapse at the UK Biobank, with "de-identified" data from its 500,000 volunteers listed for sale in China.

A message to participants in the Biobank, sent from chief executive Prof Sir Rory Collins today, notes that "listings offering access to UK Biobank data …were found on a Chinese consumer website [but] were swiftly removed before any purchases were made."

Initial reports suggest that the data was offered through Alibaba's Taobao platform, the largest online retail platform in China, which has been cited for a number of years as having problems with intellectual property (IP) protection in the US Trade Representative (USTR) annual 'Notorious Markets' report.

It is understood that the breach did not reveal any personal data, such as names, addresses, contact details, and NHS numbers, but the breach appears to have exposed data that would normally be accessible only by vetted, bona fide researchers from academic, charity, government, and commercial organisations.

According to the UK government, the breach seems to have originated from research organisations granted access to the data legitimately, rather than a hacker group, and that access by these groups has now been revoked.

Last year, news that one in five data access requests came from Chinese research groups sparked a warning from MI5, reported by The Guardian, about concerns that they may share data with Chinese intelligence agencies.

While Collins said that measures had been taken to make sure that this type of incident cannot occur again, including an upgrade that helps prevent de-identified data being taken out of the platform, the breach could dent confidence in the security of health information, just as the UK government is preparing to launch a data-sharing initiative for records held by GPs to other organisations, including the Biobank.

The Minister for Science, Innovation, and Technology, Ian Murray, told Parliament this today that at least three listings had appeared offering the data, including one that appeared to "contain data from all 500,000 UK Biobank volunteers."

He added that "additional listings offered support for applying for legitimate access to UK Biobank or analytical support for researchers who already have access to the data," and indicated that – based on discussions with the listings website owner – there had been no recorded sales of the information.

At the moment, it's not clear if the data is also being offered for sale through other channels, including the Dark Web.

A pause has been placed on access to the Biobank while a technical solution is implemented to "prevent data from its current platform from being downloaded in this way again," said Murray.

He added: "This has been an unacceptable abuse of the UK Biobank charity's data and an abuse of the trust that participants rightly expect when sharing their data for research purposes," and said the government will be issuing new guidance on control of data from research studies.

Studies conducted using Biobank data have already yielded important new findings, such as the discovery of genes that affect the risk of heart disease or cancer, new ways to predict dementia, and early warning signals for cancers and Parkinson's disease.

https://pharmaphorum.com/news/uk-biobank-patient-data-stolen-and-placed-sale-china