The Crystal That Vanished Overnight: The Ritonavir Crisis

Bölüm metni

In nineteen ninety-eight, capsules made in a drug factory from the same powder by the same method suddenly became useless. The molecule was the same, the formula was the same, the purity was the same — the only thing that had changed was the way the atoms were arranged. And the old arrangement could never again be obtained in the laboratory, by any means.

Think of the pale, greyish bloom that creeps across a bar of chocolate left too long in the refrigerator. The chocolate has not spoiled. Nothing has contaminated it. Its composition today is exactly what it was yesterday. The only thing that has changed is how the molecules of cocoa butter are arranged relative to one another. The arrangement shifted, and with it the texture changed, the gloss disappeared, the melting point in the mouth moved. The same substance, two separate solids. Chemistry's quiet sleight of hand has a name: polymorphism.

Its most familiar example is a pair everyone knows: graphite and diamond. Both are pure carbon. One is soft enough to leave a black streak on paper from the tip of a pencil; the other is the hardest natural material known. The chasm between them comes not from the identity of the atoms but from their architecture. In graphite, carbon atoms stack in layers that slide easily past one another. In diamond, every atom is bound tightly to four neighbours, building a three dimensional cage. How you pack molecules into a box determines what that box is.

In a crystal, molecules do not pile up at random. They settle into an order, a repeating pattern, and along that pattern hydrogen bonds form and lock the molecules to each other. It is entirely possible for the same molecule to settle into more than one such order, and each of those orders is called a form. The difference between forms is no academic nicety. The melting point changes, the density changes, the behaviour under pressure changes, even the colour can change. And when a medicine is involved, the most consequential of all: the solubility changes.

Solubility, for a drug, is a matter of life and death. The tablet or capsule you swallow is useful only to the extent that it can dissolve its active ingredient in the fluid of your digestive system. A molecule that does not dissolve is not absorbed; a molecule that is not absorbed never reaches the bloodstream; and a molecule that never reaches the bloodstream does nothing in the patient's body, however brilliant a discovery it was in the laboratory. The difference in solubility between two crystal forms can decide whether the same dose is an effective therapy or an expensive nothing.

Now to the molecule where that rule was tested in its most dramatic form. Ritonavir is a drug developed against HIV, blocking an enzyme called protease that the virus uses to process its own proteins. The middle of the 1990s was a period in which every month weighed heavily in HIV treatment; protease inhibitors were the missing piece of the combination therapies that, for the first time, could genuinely push back a disease nobody had been able to slow. Ritonavir was approved in the United States in March 1996 and brought to market under the name Norvir by Abbott Laboratories, or Abbott for short. It came in two forms: a 100 milligram capsule, and an oral solution carrying 80 milligrams of active ingredient per millilitre.

But this molecule had a flaw: it barely dissolved in water. Filling a capsule with the powder and handing it to a patient would mean leaving most of the molecule undissolved in the gut. To get around this, the formulation team chose an elegant but bold route: they put ritonavir into the capsule not as a solid but already dissolved. They dissolved the molecule up front in a semi solid mixture containing ethanol and oleic acid. What reached the patient's stomach was now a liquid that did not have to wait to dissolve.

That solution carried a subtle price. There is a ceiling on how much of a molecule you can dissolve in a given liquid, and in this formulation ritonavir sat just beneath that ceiling, close to the saturation limit. A solution near saturation walks a chemical tightrope. If a crystal form of lower solubility appears within it, that same solution becomes supersaturated with respect to the new form; the liquid, in other words, is now carrying more than it can hold. And such a liquid has only one thing to do: throw the excess out as a solid, and crystallise.

Through the development years, Abbott's chemists knew only one crystal form of ritonavir. In this structure, later named Form I, the molecule folded on itself in a particular way and built a particular network of hydrogen bonds with its neighbours. The ritonavir molecule is long, flexible, bendable; molecules like this can settle into different crystals by folding differently. This is called conformational polymorphism: what changes is not merely the placement of molecules relative to one another, but the posture of the molecule itself. The critical point for ritonavir was that the carbamate group in its structure could point in two different directions. When that group turns the other way, the molecule folds differently, builds a different hydrogen bonding scheme, and the result is a more tightly interlocked crystal, one that demands more energy to come apart. Ritonavir, in short, had a second face that nobody had seen. A more stable, less soluble face. And the entire architecture of the formulation rested on the assumption that no such face existed.

In mid 1998, two years after launch, a routine test in the quality control laboratory returned an unexpected result. A dissolution test measures whether a drug dissolves in a specified fluid, within a specified time, to a specified degree; it is among the most ordinary, most monotonous checks in the pharmaceutical industry. In July 1998, some batches of Norvir capsules failed it. Inside the capsules was something that should not have been there: solid particles. The solution was spitting its ritonavir back out.

Placed under a microscope and examined by X ray powder diffraction, those particles left a fingerprint in their diffraction pattern. The pattern did not belong to Form I. It was a second crystal form of ritonavir, never before seen, and it was named Form II. Its worst property emerged immediately: Form II dissolved to less than half the extent of Form I. The ethanol based formulation was wildly supersaturated with respect to this new form. The moment Form II appeared, the liquid inside the capsule was no longer a solution but a reservoir waiting to crystallise.

The real disaster came next. Once it had appeared, the new form spread within weeks across both the active ingredient production areas and the formulation areas. The reason is one of chemistry's most unnerving mechanisms: nucleation. For a crystal to form, a tiny nucleus, a seed, must first be established, and that first step is usually the hardest. If a molecule has never settled into a new form, there is no seed for that form, and a solution may carry on with its old habit for years. But once the seed exists, everything changes. A single microscopic particle, invisible, drifting in the air, settling on a bench, on the blade of a mixer, on a filter, on a technician's coat, shows the template to every saturated solution it touches. Those invisible seeds, scattered through every corner of the plant, gave the same answer to every process trying to make Form I: we make Form II now.

And the reverse proved impossible. The same reactor, the same solvent, the same temperature profile, the same personnel, the same written procedure, and the result again was Form II. The form that had been produced without incident for two years could no longer be produced at all with the recipe in hand.

Abbott spent a few million dollars on the manufacturing side trying to rescue the capsule. Different crystallisation conditions, different processing steps, cleaning strategies were tried. None of them held. In the end the company abandoned the attempt and, towards the close of 1998, the capsule was withdrawn from the market. The only option left was the oral solution, and that solution tasted terrible. At a meeting with patient representatives in October 1998, the company said it could supply enough liquid for the roughly 70,000 people using ritonavir worldwide; at the same meeting, the idea of chasing the liquid with something like chocolate or nut butter to mask the taste was discussed. And here the matter goes beyond supply logistics. A protease inhibitor regimen suppresses the virus only when it is sustained without missed days and without missed doses; otherwise resistance develops. A formulation that is hard to swallow is therefore not merely a comfort problem but a threat to the success of the treatment. On top of that, both the liquid and the capsule had to be kept refrigerated, a constraint that by itself made delivering the drug to every part of the world with equal confidence far harder.

A new capsule returned to the market in 1999. This time the formulation was rebuilt to work in spite of Form II's existence: a soft elastic capsule containing polyoxyl 35 castor oil, butylated hydroxytoluene and gelatin alongside the oleic acid and ethanol. The engineering logic had been inverted. The aim was no longer to produce the desired crystal, but to accept the existence of the undesired one and design a carrier that could live with it. In 2000, Abbott's process chemists described what they had been through in the journal Organic Process Research and Development; among the things they described was that they had ultimately been forced to learn how to make Form II deliberately and reliably, rather than by accident. The cost of the whole crisis to Abbott is usually put at around 250 million dollars.

One question remained, and it stood unanswered for years. Where did Form I go?

Crystallography has a name for this phenomenon: the disappearing polymorph. A form becomes impossible to obtain in laboratories again once a more stable rival has walked onto the stage. Sometimes a team on the other side of the world applies the old recipe and finds the new form; sometimes a team pushes in the opposite direction, trying to bring back the lost one, and fails. There is nothing supernatural about this. It is the combination of two things: nature's constant tendency to flow towards the most stable structure, and the impossibility of eliminating that structure's seeds once they exist. The examples gathered over the years by Bernstein and colleagues showed that most disappearing polymorphs had not in fact vanished; the condition that would give birth to them had merely been forgotten, or never known.

The condition that brought ritonavir's lost form back turned up somewhere no one would think to look for crystal growth: inside a mill. A study led by Aurora Cruz-Cabeza, with the participation of researchers at the University of Manchester, Durham University and in Cambridge, was published in 2024 in the Proceedings of the National Academy of Sciences. The instrument the team used was a laboratory scale ball mill: a steel vessel, steel balls inside it, and a motor that flings the vessel back and forth 30 times a second. The powder is broken and reshaped over and over between the balls. Add a very small amount of liquid, a wetness on the order of a quarter of a microlitre per milligram, and a thin film forms that allows molecules to rearrange on the surfaces.

The results ran in two directions, and both were striking. Milled with a drop of isopropyl alcohol, Form I converted to Form II within just a few minutes. The form that sealed off a factory in 1998 is, today, something you could find in a laboratory before lunch. The second direction was the more surprising one: with a drop of water, at room temperature, after about 150 minutes of milling, Form II converted completely back to Form I. The vanished crystal had not vanished.

The reason is not an exception to thermodynamics but the fact that thermodynamics is sensitive to scale. When we calculate a crystal's stability we usually look at the energy of its interior, because in a large crystal the molecules sitting on the surface are a negligible fraction of the total. In the mill that balance breaks down. Once crystals are broken down to the order of 40 nanometres, a substantial share of the molecules begins to live on the surface, and surface energy becomes the decisive term in the arithmetic. The shape of the crystal joins in too; needle like elongated structures carry a great deal of surface relative to their volume, and so amplify the effect. Add to this the molecule's ability to change its own fold, and the stability ranking between the two forms can be reversed. Form II, the more stable one under normal conditions, is no longer the winning side at the nanometre scale with the right degree of wetness. The mill opens a new dial for the polymorph hunter: by changing the liquid, the duration and the ratio of wetness, you can deliberately tilt the stability ranking.

Something similar matured on the computational side. A study published in 2025 in the journal Communications Chemistry by researchers at Merck set out to predict ritonavir's crystal structures from scratch on a computer. The method, called crystal structure prediction, generates every possible packing arrangement of the molecule and ranks them by energy. The result was unambiguous: around room temperature, the structure predicted to be most stable was the one corresponding to Form II, sitting about 1.62 kilojoules per mole below Form I. That is consistent with the measured difference, which falls in the range of 2 to 3 kilojoules. The authors' judgement is clear: had today's computational tools been available at the time, the existence of Form II and the formulation's intolerance of a drop in solubility could have been foreseen. The same study also solved, using 3 dimensional electron diffraction, the structure of a fourth form of ritonavir: needle shaped, carrying a disordered isopropyl group inside it, and markedly less stable than the other two forms.

Ritonavir's own story, meanwhile, arrived somewhere else. The tablet formulation approved in 2010 follows a path that solves the polymorph problem at the root: hot melt extrusion. The active ingredient is melted and mixed together with a polymer, and what emerges is an amorphous solid dispersion that has never crystallised at all. Where there is no crystal there is no polymorph; what remains is to prevent the amorphous structure from crystallising over time, which is precisely the polymer's job. The tablet can be stored at room temperature and Form II does not appear inside it. The cold chain constraint thereby passed into history. The molecule itself found a second career: ritonavir today is mostly used not to suppress HIV directly but as a booster, slowing an enzyme in the liver so that the drug taken alongside it stays in the blood longer. The second component of the nirmatrelvir and ritonavir combination used in Covid 19 treatment is the same molecule.

What stopped a factory in 1998 was not a virus, not sabotage, not a calculation error. It was a second way of folding, belonging to the same molecule, that nobody had seen. Finding that form takes half a day in a laboratory today; and the old crystal that its seeds took from us comes back inside a steel vessel, milled with the right liquid for the right length of time.

Nöron Science'de bugünkü bölümün sonuna geldik. Yeni bir hikâyede yeniden buluşmak üzere.