DBEC: How Old Are Your Knees?
You think you get unusual mail? Consider Thayer School professor John Collier's mailbox.
For 20 years, surgeons have been sending him used knee- and hip-replacement prostheses, retrieved at the time of failure or death. But this is no backward-looking enterprise. Recently, findings from this unique "implant-retrieval" lab have set the knee- and hip-replacement world abuzz.
Before 1976, joint replacements were nearly always cemented in place with polymethylmethacrylate (aka: Plexiglas®). Then research done by Collier, Dartmouth Medical School professor of clinical surgery Michael Mayor, and others led to the development of the first non-cemented prostheses, which Mayor was among the first clinicians to test. Today, nearly half the 300,000 knee and hip prostheses implanted in the U.S. each year are anchored with the aid of a porous coating of cobalt alloy or titanium alloy beads, which encourages in-growth of the patient's own bone tissue, creating a fixation system the body can maintain itself.
The focus of Mayor and Collier's 20 years of implant-related study is extending the life and improving the performance of future generations of such hardware. For the past 16 years, they have worked closely with Depuy Inc., of Warsaw, Indiana, one of the principal manufacturers of non-cemented orthopedic implants, and with Depuy-DuPont Orthopaedics, a joint venture of Depuy and DuPont.
"Implant manufacturers can't do this kind of work themselves," claims Collier. "The average company gets back very few knee or hip prostheses, and generally, only its own. And almost all have constraints on them preventing them from being sectioned and tested-in many cases they're evidence."
The Collier-Mayor salvage operation is a booming business:
"In the early days, we looked at two or three, maybe five implants a year. But as the number of devices that were out in the community and the recognition that we were doing this increased, our numbers have gone up to about 600 a year. To date, we've looked at about 4500 of them." What they have learned is astonishing.
Even bionic knees wear out
"From laboratory testing of wear rates of the materials used, you should be able to get between 20 and 30 years out of these devices," says Collier. "Yet their average lifetime is probably closer to ten. You'd hope the wear rate would be on the order of 100 microns (0.1 mm) a year. Yet we're often seeing rates much closer to a millimeter a year - ten times what you'd expect."
Failure at the anchoring surface is no longer the problem. Instead, early retrieval-lab studies showed the implants' polyethylene bearing inserts were wearing in unusual and accelerated ways - ways not obviously attributable to differences in design, stress, or installation.
"Each manufacturer's product has a different design," allows Collier. "Knee prostheses, for example, vary in how the femoral component matches up with the tibial component. But when you measure the contact stress between them, they vary by only a factor of two, the one with the lowest contact stress having about half that of the highest.
"Or, you might expect wear to be related to high loading conditions - as in a heavy patient who is very active. But, you can get two devices that look very similar, implanted in similar patients, where one has failed and one hasn't. "Even more puzzling is that you'd expect wear to begin at the insert's surface - but instead, the material comes apart from within, even though the insert was cut from a uniform block of polyethylene."
Collier describes the damage graphically: "It's like if you have a plastic bag full of ice and you smash on it with a mallet, the bag may still be intact, but the ice below starts breaking into little particles. Finally you hit a hole in the bag and all the stuff comes out." Something other than design, stress, surface anomalies, or installation technique was causing the accelerated material failure. But what?
Enter the Dream Team
To solve the mystery, Collier and Mayor and their sharp, supportive, and highly motivated research group - comprising engineers, technical staff, and students - began to focus collaboratively on one aspect of the problem at a time. Collier describes his team in materials-analysis terms.
"There's a lot of good group dynamics, a lot of laughing and kidding, and no backbiting. There's almost no emotional energy being lost between these people."
Among the team, comprising two husband-wife pairs, several Thayer School master's and doctoral degree candidates, and a skilled staff, both the work and the credit are widely shared; an impression particularly apparent last winter, just before they were to travel to Orlando to deliver an award-winning paper at the annual meeting of The Hip Society (a subset of the American Academy of Orthopedic Surgeons).
"Two of our researchers, recently graduated Lauren Sutula (M.S. '94) and doctoral candidate Dan Sperling, gave presentations at the meeting," Collier said proudly. "After all, it was Lauren's finding that got us there."
But the mystery tale began much earlier.
"About four years ago, we began focusing exclusively on figuring out what was happening to the polyethylene. And in the last two years, the problem began to move for us."
Seeing spots: the first clue
One day in the summer of 1992, Mayor was examining an acetabular cup, the shell-like polyethylene insert which serves as a sort of "bearing" for the hip prosthesis.
"He put light behind one," says Collier, "and suddenly noticed all these black spots in it. He looked at a bunch of them, and a lot had black spots. We started looking at knee inserts, and they had the black spots too!"
Searching the literature and phoning manufacturers yielded only a name for the spots: "fusion defects." Manufacturers, though intrigued, hadn't a clue where they came from.
Intensive testing of bar stock samples showed the spots to be polyethylene particles not measurably dissimilar from the bulk material, but which had not bound to it for some reason. Evidently, prosthetic-grade polyethylene was not a totally homogeneous material, even though almost all comes from powder supplied by a single manufacturer.
Collier's team began to run the cups through a microtome, making sections a fifth of a millimeter (200 microns) thick, back lighting them, and counting the defects, using computer-assisted imaging techniques.
"When we looked at stock produced recently," says Collier, "we found no defects. Yet when we looked at retrieved implants - and we went through our whole database of hundreds of them - it turned out about 65% had them.
"We had some hypotheses: maybe the body's warm, fluid environment caused changes in the material over time, although we could think of no mechanism. In any case, that wouldn't explain why inserts that had never been implanted also had the black spots!
"Maybe they occurred only in older stock. Maybe they were a function of the production process, occurring cyclically, due to one shift doing something differently from another."
To test the latter hypothesis, Collier and company went so far as to do on-site quality assurance for one of the large manufacturers for an entire run of material - several thousand feet - checking it every 10 feet. They even analyzed material from the company's archives. At first they found almost no defects, and then ...
"All of a sudden, we found seven separate lots of bar stock from one company that all had defects. Tracing it back, all seven came from the same batch of powder, powder from one source. But, what about this powder was different?"
Normally, polyethylene has extraordinary ductility and strength; it can stretch to two to ten times its length and still maintain its strength. However, this defect-ridden lot demonstrated a 20% reduction in mechanical properties, compared to specimens from other lots.
"Obviously the defects mattered," says Collier succinctly.
However, subsequent wear and fatigue tests done using knee and hip simulators failed to replicate the fatigue propagation seen in the returned implants.
"In the failed knees," says Collier, "it looked like the material all of a sudden had changed characteristics and become like glass - in essence, it was shattering. None of our test specimens did that."
Equally puzzling was the fact that statistical analysis showed a "modestly good" correlation between defective stock and damage in knees, but not in hips. ("Damage" included things like cracking, pitting, delamination, burnishing, abrasion, and "creep," or displacement of the material under stress.) Nor was it a one-to-one correlation; there were inserts that had lots of defects and no damage, and inserts with lots of damage and no defects.
Behind the white line
Then, back in the Thayer School lab, the team made a key observation: every microtome section from defective inserts showed a thin white line just a millimeter inside the perimeter and following its every contour. Microscopic analysis showed it to contain hundreds of tiny light-refracting cracks. How did the white line's presence correlate with the severity of damage?
"One-to-one," says Collier. "If the knee insert showed lots of damage, it invariably had one of these white lines. Especially if it had cracking and delamination, the most disastrous flaws because they mean the material is literally coming apart."
Further, Collier's colleagues at Depuy-DuPont Orthopaedics determined that the white line showed very heavy oxidation, although the surface did not.
Maybe something in the fluid environment of the knee joint was attacking the polyethylene, or maybe stress was pumping oxygen into the insert's slightly porous surface, causing oxidation over time. Several hypotheses were suggested, but none panned out in simulation testing.
"Then one day last year, we got very , very lucky."
A plastic "Rosetta stone"
One day in June, 1993, Lauren Sutula, who was working with Collier at the time, stumbled on what Collier likes to call "the Rosetta stone for polyethylene."
"She came across two inserts sitting on the shelf - both 14 years old, still in their original boxes, both from the same manufacturer, both unopened, and - we learned later - both from the same lot of polyethylene. Everything about these inserts was similar until she cut them. One had no white line, but the other had one so severe and so thick that the sections of the insert actually came apart in her hands!
"Here were inserts that have never been implanted, never been stressed, never been loaded, the boxes had never even been opened - and one falls apart and the other doesn't!"
The only noticeable difference was a label notation saying the brittle one had been sterilized and the other hadn't. (Since about 1980, the sterilization method of choice has been gamma-ray sterilization in air.) So the team went back through their samples again, sectioning some from the early 1970s that they knew had not been gamma-sterilized.
The results: no white line and almost no damage. They also found a one-to-one correlation between damage and the white line, not just for hips ( "which crack and delaminate only about 15% of the time") but for knees, too; where there was damage, there was always a white line.
"Obviously gamma sterilization, the white line, and the damage were correlated," says Collier. "But how? Why? And what else was happening?" The team moved into a higher gear and sharper focus. Developing a technique for studying the tensile strength of the microsections, they found the normal ductility and tensile strength of the polyethylene in flawed knee and hip insert specimens was reduced by about 50%.
"Even more interesting," says Collier, "was the solid correlation between age and damage. Inserts over three years old often showed delamination and cracking, and when they did, they always showed the white line. Those less than three years old very rarely showed cracking and delamination, and - interestingly - never showed the distinct white line!
"In fact, if you take a piece of polyethylene and subject it to gamma-radiation sterilization, then take it right back into the lab and test it, the properties are nearly the same as before you sterilized it; it doesn't change much. But if you let that piece sit on the shelf for a year after sterilization and then test it, the properties are down. Two years later they're down pretty significantly. And three to four years later, the material is sufficiently brittle that when you cut it with your microtome, you actually create the white line and all those micro-cracks."
How old is your polyethylene?
Collier believes gamma-radiation sterilization breaks up polymer chains, creating free radicals. Done in air, this permits oxygen to bind to the breaks in the chains, oxidizing more and more of them as time goes on. Why the surface remains unchanged is unclear, although it may involve competing reactions: one at the surface, and one below.
In order to test the insert's three zones independently - the clear surface, the white line inside, and the center of the insert - the team began making cuts parallel to the insert's surface rather than straight through its thickness. What did they find?
"The material on the outside maintained almost all its original properties; the material in the white band showed up to 90% reduced strength and ductility; and in the center - depending on how long ago it was sterilized - the material was actually brittle."
"Obviously when an insert was sterilized is critically important. Regardless of whether an insert is implanted or sitting on the shelf, its mechanical properties appear to be going downhill at about the same rate."
This was clearly earth-shaking new information. What should be done with it?
And what if the team were wrong? What if the delamination and cracking were due to something else they hadn't yet uncovered? Although there was no good reason to believe manufacturers weren't telling them the truth, admittedly the team was operating on second-hand information; they couldn't say the phenomenon was related exclusively to gamma sterilization in air without knowing every detail of the manufacturing process.
"To do that, we'd need access to their entire production process, which they obviously wouldn't allow. But what were our alternatives - to put some brand-new sterilized and unsterilized inserts on the shelf, and/or implant them, and wait three years to check them for damage? Who could wait that long?"
Collier's longtime colleague, research engineer Vic Surprenant, set about finding some way of accelerating the damage-development process in the lab, and ended up creating a sort of simulated physiological "hypertime."
"Instead of subjecting the inserts to a normal body environment of 37 degrees Centigrade," says Surprenant, "you increase the temperature to 80 degrees. And instead of 20% oxygen, as in normal atmosphere, you go to 100% oxygen. If you do both of those things for 21 days, you've got an environment that should accelerate the deterioration."
Why 21 days? Collier chuckles: "Vic has this ability to make wild guesses and hit the mark; it's been his habit for years, and drives the rest of us crazy."
Sure enough, after 3 weeks in the accelerating environment, all the specimens showed white lines, oxidation effects, and decreased mechanical properties. The team couldn't say what else manufacturers might be doing to exacerbate or delay the effect, but they could say that these conditions alone would produce the phenomenon.
Said Collier, "If you take a standard piece of polyethylene, if you subject it to gamma sterilization in air, and if you wait ... this phenomenon will occur. It won't occur 100% of the time at this same rate - that is, different variables will delay or accelerate it. But if you're patient, it will happen."
Influencing the industry
The team figured once the industry learned of their findings, they would have a number of alternatives.
One was to go back to old-fashioned ethylene oxide sterilization, a method abandoned years ago because of reduced efficacy, toxicity, and residual surface contamination.
Another was to halve the gamma-radiation dose, which - by similar testing - they found also reduced the damage, or at least increased the time required for the prostheses to show significantly reduced mechanical properties.
Thirdly, they could sterilize in a vacuum, or in nitrogen or argon. While this would still create free radicals, it would prevent the dreaded oxidation and subsurface cracking - at least until the day the package was opened.
Combining these findings, Collier predicted that by using low-dose gamma radiation in an inert atmosphere and packaging it so that oxygen was kept from the prosthesis, the average useful life of an implant would be measurably increased.
Not content to rest on partial improvements, however, the team is beginning to test a sterilization technique currently used in Canada, involving creation of ionic species within a gas plasma to kill microbes on implant surfaces.
"Sterilizing them with gas plasma or ethylene oxide should eliminate the severe oxidation, reducing the frequency of cracking and delamination in the polyethylene prostheses and thereby raising the potential for greater numbers of components to provide 20 or more years of useful life," says Collier. "And so far we've found no deleterious effects. Our guess is this'll be the direction a good chunk of the manufacturers take."
Ethylene oxide is currently available, but the gas plasma technique is still pending FDA approval. The approval requires proven sterilization effectiveness with no negative effects on implants, something Collier expects will occur within a year. In the meantime, could ways be developed to screen existing inserts for deterioration before installing them in patients?
Collier believes it's better to encourage the industry to switch immediately to different sterilization methods. But what if, while the team is working to extend the useful life of polyethylene inserts, another material comes along to make polyethylene a thing of the past?
"Polyethylene's been around since the late 1960s," says Collier. "I know there are people studying ceramic-on-ceramic and metal-on-metal inserts, because they think that's the way to solve this articulation problem. But no work I've seen supports metal-on-metal or ceramic-on-ceramic surfaces in the knee. In the hip, where you have a ball and socket, either one might. But I think they'll be a lot less forgiving than polyethylene, and much more expensive."
How fresh is your hardware?
Modified sterilization won't be the only change resulting from the team's work; packaging will change too. For the first time in the history of orthopedic hardware, freshness will be a factor.
"Right now, the polyethylene components often have no serial numbers, no identification, no sterilization information or dates on them. So, as a surgeon, you can't walk up to an O.R. stock clerk and say, - I want the most recent one that's been sterilized.' In the future, implant packaging may be far more informative than it is today; it'll say how and when it was sterilized, and if there's any relation between that and shelf-life. If it's gamma sterilized, even at low dose, it'll be packaged in an inert atmosphere, so the clock won't really start ticking until you open the package. And packaging will be tested to see that it actually keeps oxygen out.
"Of course, if everyone goes to ethylene oxide or gas plasma sterilization, this information will be less critical but still nice to know. Why get something sterilized five or six years ago, knowing that, even if it oxidizes very slowly, it's already different from one sterilized just yesterday?"
The whole world is watching
To Collier's knowledge nobody else has duplicated his team's award-winning work, which will doubtless affect hundreds of thousands of lives.
"We're talking about doubling or tripling the lifetime of the average implant," he says emphatically. "The old idea - no matter what you have it's probably going to have to be replaced in your lifetime - may go out the window. Maybe we can make these articulations last for your lifetime."
He may also have eliminated another huge problem with existing prostheses. As polyethylene deteriorates, it releases debris into the joint - tiny polyethylene particles that trigger a foreign-body rejection reaction that often leads to bone loss at the implant site.
"Bone cells try to destroy the particulates by eating them, but since polyethylene won't dissolve, the bone cells die. This releases osteolytic enzymes that dissolve bone, creating a cascade effect that eventually loosens the prosthesis. The congruency of the opposing surfaces gets poorer and poorer until ultimately the prosthesis fails completely."
Collier is confident that once this news gets around, doctors will start following their implant patients' x-rays more closely post-operatively, and as soon as they see a serious mismatch in the gap between the implant surfaces - the standard indication of insert wear - they will consider replacing the inserts with new, differently sterilized ones, rather than wait as they do now until the inserts wear out. That way there will be less debris to worry about, and the implants won't have time to work loose.
"So it's not as if everyone getting a knee replacement today is suddenly up the creek. If the polyethylene bearing wears out, it can be replaced, this time with material which should be more resistant to oxidation and less likely to demonstrate reduced mechanical properties over time.
Perhaps when it's repaired, it'll be repaired for life, because the new stuff won't develop the problems that the old stuff did." What's next?
When the dust settles, they would like to begin exploring more subtle mechanisms for improving polyethylene inserts, "like how to get from 0.1 mm wear in a year to 0.05 mm. Then we can begin talking about ultra-smooth surfaces and other things that are difficult to even consider now. Right now the potential change in expected performance with the change in sterilizing technique is so big, it dwarfs everything else."
From John Collier . . . A doctor friend in Boston told me an interesting story. A knee he put in a patient 9 years ago worked great for 8 years and then failed very rapidly. When he asked for the patient's chart, there in the file was a sealed box with another sterile polyethylene insert in it; for some reason, he'd been given two at the time of the original surgery. So - because the knee was of an older design and it wasn't easy to get new parts for it - he was delighted to just pop in this other insert.
Nine months later, the patient had worn clear through it!
Why? Because the one was oxidizing just as fast, sitting in the box, as the other one was, in the patient. This one failed real fast because it couldn't resist the load put on it as well as the first one had. The first one lasted 9 years - and, really, the second one lasted 9 years too, but 8 years of that was sitting on the shelf and just 1 year was in the body! So you can see that it's the time after sterilization that matters more than anything else.
Free-lance writer Mary vanRoden was at first uneasy about doing this article, as she herself had both knees replaced (at DHMC) in May, 1992. Although "they are functioning just fine, thank you" and represent "a whole new lease on life," she admits continuing concern. "However," says she, "it must be better to know what's going to happen, than to be blind-sided by it down the pike. Mustn't it?"