Gadolinium Contrast Safety: What the ACR Group I, II, and III Labels Actually Tell You

The short answer#

The American College of Radiology (ACR) does not rank gadolinium contrast agents by how their molecules are built. It ranks them by how many people they have actually harmed. The Group I, II, and III labels in the ACR Manual on Contrast Media count confirmed cases of nephrogenic systemic fibrosis (NSF) tied to each agent, so a label is really an epidemiologic verdict resting on a chemistry story rather than the chemistry story itself. Group I holds the agents behind most historical NSF cases and has essentially disappeared from US practice. Group II holds the agents with few, if any, clear-cut cases and does most of the work in imaging today. Group III holds the handful of agents whose data are still too thin to call.

Key points#

• The ACR groups sort agents by observed NSF cases, specifically "unconfounded" cases where one agent can be blamed without a second agent clouding the picture.
• Group I agents are contraindicated in the highest-risk kidney patients and are no longer in routine US use.
• Group II agents may be given even to patients on dialysis or with advanced kidney disease when needed, at a risk expert panels call exceedingly low.
• "Group II" is not the same as "macrocyclic." One linear agent sits in Group II on the strength of its safety record.
• Tissue retention of gadolinium is a separate question from NSF, and the group label does not answer it.

Start with the axis that chemistry sits on#

Every gadolinium agent is the same basic object: a reactive metal ion that would be toxic on its own, locked inside a carrier molecule called a chelate. The whole safety question turns on how well that lock holds.

Two properties describe the lock. Thermodynamic stability is how firmly the ion is bound when everything sits at equilibrium. Kinetic inertness is how slowly the ion can wriggle free once conditions change. The structural shape that governs both is whether the chelate is macrocyclic or linear. A macrocyclic agent surrounds the ion inside a closed ring, like a hand fully cupped around a marble. A linear agent uses an open chain that can loosen its grip. Macrocyclic agents are markedly more inert, and that difference is the leading mechanistic reason a group of older linear agents caused most of the trouble.

Hold that axis in mind, because the clinical labels line up with it only partly.

The disease that forced the labels#

NSF is rare and can be disabling. It thickens and hardens the skin and connective tissue and can reach internal organs. Clinicians first described it in the late 1990s, and by 2006 investigators had linked it to gadolinium in people with severe kidney impairment. The favored mechanism is that some gadolinium separates from its chelate and settles in tissue, where it appears to drive fibrosis. Almost every confirmed case occurred in patients with advanced kidney disease, acute kidney injury, or dialysis dependence. The common thread is slow clearance: a healthy kidney flushes the agent in minutes, while a failing one lets it circulate for far longer, giving the metal time to escape.

As the case reports piled up, the ACR needed a way to convert scattered signals into practical guidance. The three-group system is that translation. The groups count unconfounded cases, meaning episodes where one agent can be held responsible without another gadolinium agent muddying who did what.

Reading the three groups#

Group I: the agents that left#

Group I holds the agents behind the largest share of NSF cases: gadodiamide (Omniscan), gadopentetate dimeglumine (Magnevist), and gadoversetamide (OptiMARK). The FDA contraindicates them in the highest-risk patients, such as those with acute kidney injury or chronic stage 4 to 5 kidney disease. In practice they have dropped out of routine US imaging.

Group II: today's workhorses#

Group II holds the agents with few, if any, unconfounded NSF cases: gadobenate dimeglumine (MultiHance), gadobutrol (Gadavist), gadoterate meglumine (Dotarem and Clariscan), and gadoteridol (ProHance). In 2023 the ACR added gadopiclenol (Elucirem, Vueway) after judging its stability comparable to the rest.

The practical payoff is large. For standard or lower doses of a Group II agent, the ACR treats screening kidney function by questionnaire or lab test before injection as optional, because any residual NSF risk is low enough to be possibly nonexistent. A joint ACR and National Kidney Foundation consensus went further: a Group II agent may be given when clinically needed even to a patient on dialysis or with stage 4 to 5 disease, at a risk the panel called exceedingly low.

Group III: still thin#

Group III is the small, unsettled category for agents with limited NSF data but no clear pattern of unconfounded cases. Gadoxetate disodium (Eovist), a liver-specific agent, sat here for years until it gathered enough safety data to be handled alongside Group II in the most recent manual.

The trap: Group II is not a synonym for macrocyclic#

This is where the chemistry axis and the clinical label part ways, and where quick summaries go wrong.

If chemistry alone decided the label, every Group II agent would be macrocyclic and every linear agent would sit lower. That is not what happened. Gadobenate dimeglumine is a linear ionic agent, and it still earns a Group II spot because its observed safety record put it there. The takeaway is that the ACR groups report outcomes while the macrocyclic versus linear split explains a mechanism. A careful reader uses both and refuses to assume the molecule's structure predicts every clinical result.

Retention is a different question than NSF#

A separate worry surfaced around 2017. The FDA reported that trace gadolinium lingers in tissues, including the brain and bone, for months to years after injection, and that linear agents leave behind more than macrocyclic ones. The same agency stated it had found no adverse health effects from this retention in patients with normal kidney function, and that the benefit of these agents kept outweighing the risk. It then required a class warning and a patient Medication Guide across all gadolinium agents.

Reading the evidence well means keeping two questions apart. That gadolinium is retained is a measurable fact. That retention harms people with healthy kidneys is not established. The group classification answers the NSF question and only that question. Treating a Group II label as a verdict on every gadolinium debate, including the contested cluster of symptoms some patients attribute to deposition, reads more into the label than it holds.

How to read a group label like a regulator#

Pull the threads together and a simple habit falls out. A group label is outcome-driven, it hinges on unconfounded cases, and it moves as agents accumulate data, which is exactly what gadopiclenol and gadoxetate show. So when you see one, read it as a running summary of what has been observed under real dosing, cross-checked against a plausible chemical mechanism, not as a fixed property stamped on the molecule.