Samples of Cladosporium species (L), and Aspergillus fumigatus, two of the fungi diagnosed in the fungal meningitis outbreak . (Harrison McClary/Reuters)

The drug that has caused 696 fungal infections and 45 deaths over the last six months isn’t one you want sloppy amateurs making.

A long-lasting anti-inflammatory, it is difficult to manufacture in sterile form. Injected into joints or around the spine to relieve pain, it stays in the body for months. It has a bad side effect of suppressing defenses — a real problem if it’s contaminated. If the contaminant is a microscopic fungus, the drug feeds it like fertilizer.

In short, methylprednisolone acetate — an “injectable steroid,” in pharmaceutical parlance — is a prescription for disaster if not made right. A few drug companies make it. So do an unknown number of “compounding pharmacies,” a little-known, lightly regulated shadow world of the American pharmaceutical industry.

And that, recent history suggests, is a problem.

One company, New England Compounding Center (NECC) of Framingham, Mass., made tens of thousands of doses of methylprednisolone acetate and two other steroids (betamethasone and triamcinolone) used for similar purposes. It sold them to pain clinics in 23 states, satisfying a demand for preservative-free medicines not available from mainstream drug companies. Last year, samples of all three products — including the deadly lots of methylprednisolone — were found to be contaminated with fungus or bacteria.

Here are basic steps a compounder could follow to make a drug.

The dangers of injectable steroids made by compounding pharmacies have been clear for a while.

In 2001, 11 people in Northern California developed bacterial infections from injections of the steroid betamethasone made by a compounding pharmacy. Three died. A year later, two people in Michigan developed bacterial meningitis from methylprednisolone; neither died. That same year, seven people became ill and two died from fungal infections caused by contaminated methylprednisolone made by a compounding pharmacy in South Carolina. Last spring, contaminated steroids injected into the eyes of patients in Florida caused 13 vision-damaging infections.

Hard to make and dangerous to use, injectable steroids are an example of how small risks can converge to form a big one.

Many challenges

How NECC’ s methylprednisolone acetate became contaminated with Exserohilum rostratum — the fungal microbe responsible for most of the deaths — isn’t known. The company isn’t answering questions. But industry experts and FDA officials said there are many challenges to making and keeping the drug sterile.

Methylprednisolone acetate is sold to doctors in single-dose glass vials with rubber stoppers and a metal foil cap. It is a suspension in water, a powder not fully dissolved that turns the liquid cloudy when the vial is shaken.

The most reliable way of ensuring that a liquid drug is microbe-free is “terminal sterilization.” That means heating the final product hot enough and long enough to kill anything in it. Unfortunately, that can’t be done with methylprednisolone acetate because the substance would break down.

An alternative is to pass the liquid drug through a filter that strains out bacteria and fungi. But that’s not possible either, because it would catch the drug crystals as well as the contaminants. Drugs can also be sterilized with a gas called ethylene oxide and with ionizing radiation. But it’s impossible to use the gas on the final product, and radiation damages rubber stoppers.

As a consequence, the only way to make injectable methylprednisolone acetate safely is to sterilize the components — and then keep them from getting contaminated during assembly.

Three companies are licensed by the Food and Drug Administration to make methylprednisolone acetate. (The FDA doesn’t license compounders, and the agency generally isn’t involved in their operation unless serious problems arise.) Pfizer sells a branded version called Depo-Medrol, and Sandoz and Teva have generic versions. Sandoz would not answer questions about production and Teva didn’t answer phone or e-mail messages at all. Pfizer agreed to provide a general description of how the company makes its drug.

Pfizer concocts its own methylprednisolone rather than buying it as an “active pharmaceutical ingredient” (API) from a supplier. The chemical is synthesized in a clean plant, which keeps the microbial contamination (“bio-burden”) low. It’s then milled into powder, put in pouches and exposed to sterilizing gas.

Pfizer’s steroid is already sterile by the time it gets to the assembly line, where it’s mixed with sterile ingredients and packaged in sterile containers. Air flow, airborne particulate counts and other environmental conditions are monitored with each production run. All vials are inspected. Samples are opened and tested.

“It’s the overall operating culture that ensures the sterility and potency of every product we are manufacturing,” said Niall Condon, a Pfizer executive.

Compounders probably get their API from suppliers, many of which are in China and India. The material is unsterile and occasionally grossly contaminated. Whether the compounders sterilize it in solid form by heat, gas or radiation, or dissolve and recrystallize it (another strategy) isn’t known. But once that is accomplished, there remain many opportunities for contamination.

FDA inspectors who visited NECC’s compounding pharmacy found gross contamination on work surfaces. The company turned off the air handling system at night. The building was located next to a dirty recycling center.

“The risks involved in doing ‘aseptic compounding’ can be controlled,” said Harold Baseman, a former drug company official who runs the consulting firm ValSource, in Downingtown, Pa. “The way to do it is training, proper design of the facilities, and procedures. It’s important to do it right.”

Averse to preservatives

Pain doctors turned to compounding pharmacies for injectable steroids for several reasons. One is supply. The three steroids used for back and joint pain have periodically been subject to shortage.

Sandoz and Teva currently aren’t making their versions of methylprednisolone acetate. (Neither would say why not.) Pfizer, which does continue to make it, is apparently able to satisfy demand, as the drug isn’t on FDA’s list of pharmaceutical products in short supply.

Compounded steroids are also slightly cheaper than ones made by drug companies. The big draw, however, is that they don’t contain preservatives.

“I think you would be hard-pressed to find a physician who injects a preservative-containing compound. Everybody just shies away from it,” said Christian Gonzalez, a Florida anesthesiologist who is vice president of the American Academy of Pain Management.

All of the FDA-licensed steroids contain one or more chemicals added to kill microbes. Their purpose isn’t to backstop the manufacturing process. Drugs in sealed vials and labeled “sterile” are assumed to be free of microbes even if they don’t contain preservatives.

Instead, preservatives are designed to protect against contamination when a doctor or nurse sticks a needle into a vial holding multiple doses of medicine. The FDA requires preservatives in multi-dose vials. It prefers that single-dose vials not contain preservatives. However, for historical reasons drug companies also put preservatives into their single-dose vials of steroids (which is how nearly all the products are sold).

Companies could apply to the FDA to sell preservative-free methylprednisolone. For some reason, none have. The entire market has been left to the compounders.

Many physicians using injectable steroids don’t want preservatives because of some evidence the additives — especially one called benzyl alcohol — can irritate nerves, kill cells in the retina, and damage the brain of premature babies. Labels on the FDA-approved drugs mention the hazards.

How relevant those worries are for injections into the “epidural space” outside the spinal cord is uncertain. But because a needle could accidentally go in the wrong place, most doctors have steered clear of preservatives.

“It is a general consensus not to use substances with preservatives, specifically, toxic ones, into the epidural space. That is the traditional teaching,” said Laxmaiah Manchikanti, a physician who heads the American Society of Interventional Pain Physicians.

Nevertheless, some doctors are leery about the preservative-free alternatives.

“I was very close to going with a compounding pharmacy,” said Scott Glaser, a 52-year-old pain physician in Chicago, recalling a sales pitch he got at a medical conference a few years ago. He stuck with what he was using: drugs containing preservatives. “If anything ever happens, I would just feel better if I’m using stuff made by a national producer.”

A paralyzed immune system

The painkilling steroids are man-made variations on cortisol, a hormone made by the adrenal gland. Cortisol has many functions. One of them is suppression of the immune response, the body’s ability to kill microbial invaders. That makes contaminated steroids especially dangerous.

“One of the reasons why so many patients got infected is that their normal defense mechanism was bypassed,” said Stuart M. Levitz, an expert in fungal infections at the University of Massachusetts Medical School.

David W. Denning, a researcher at the University of Manchester in England, concurred.

“It’s highly likely the immune system is effectively paralyzed in the immediate vicinity of the injection,” he said.

Deep fungal infections are rare and hard to cure. The organisms grow slowly but relentlessly, penetrating blood vessels and forming abscesses. To cause meningitis, they must penetrate three layers of tissue protecting the spinal cord.

To make matters worse, the fungus in NECC’s steroid may have been fed and watered by the drug it was contaminating.

In 1994, Denning and two collaborators published a study showing that cortisol boosted by 40 percent the growth rate of two species of Aspergillus fungus. The experiments were done on laboratory culture plates; the results had nothing to do with the suppression of immunity.

In short, fungus in steroids was fungus on steroids.

Aspergillus is the most common cause of fatal fungal infections. However, little is known about E. rostratum, the species responsible for nearly all the recent deaths. Could its growth also be revved up by steroids?

“I wouldn’t like to guess,” said Denning, who co-led the international project to sequence the Aspergillus genome in 2005. “We don’t understand the mechanism. It’s a biological loose end.”

Nobody has studied the fungus-on-steroids problem in two decades. It wasn’t worth the effort. But it may be now.