In this unfair competition case under 19 U.S.C. Sec. 1337 (1982), appellants American Hospital Supply Corporation and Massachusetts General Hospital (American) appeal from the determination of the U.S. International Trade Commission (Commission), that appellees Travenol Laboratories, Inc., and Pfrimmer & Company (Travenol) are not in violation of section 1337 because they do not infringe Fischer, U.S. patent No. 3,950,529 (the Fischer '529 patent). We affirm.

The Fischer '529 patent, which is assigned to American, relates to an amino acid nutritional formulation for patients with liver disease, and a method of using that formulation. The formulation is a mixture of amino acids falling within a specific range of proportions. The formulation can be administered either intravenously or orally.

Liver-diseased patients need to receive sufficient protein for adequate nutrition, yet, these patients sometimes develop intolerance to protein which can result in hepatic encephalopathy,1 which can in turn lead to coma or death. By restricting protein intake, the tendency of a liver-diseased patient to develop hepatic encephalopathy can be reduced; however, limitation of protein intake can lead to malnutrition. The claimed Fischer formulations are intended to provide adequate nutritional support while reducing the incidence or severity of hepatic encephalopathy. They are also used as a treatment for hepatic encephalopathy.

Claim 1 of the Fischer '529 patent is representative of the claims alleged to be infringed:

An amino acid formulation for administration to human patients with liver disease, comprising a mixture of the following essential and nonessential amino acids combined in proportions defined by the following interrelated molar ranges:

Amino Acids      Molar Ranges
---------------------------------
L-isoleucine     0.549-0.0823
L-leucine        0.0670-0101
L-valine         0.0574-0.0861
L-tryptophan     0.000816-0.00441
L-phenylalanine  0-M
L-tyrosine       0-0.00300
L-lysine         0.0333-0.0500
L-methionine     0.00491-0.0147
L-threonine      0.0228-0.0454
L-alanine        0.0686-0.103
L-arginine       0.0275-0.0413
L-histidine      0.0124-0.0186
L-proline        0.0556-0.00834
L-serine         0.0152-0.0571
glycine          0.0451-0.144
L-aspartic acid  0.0451
L-glutamic acid  0.0702
L-ornithine      0-0.0382
L-cysteine       0-0.00228
---------------------------------

wherein M represents the upper limit of the range for phenylalanine and is equal to 0.009 minus the respective molar amount of tyrosine present in said mixture, the combined molar amounts of phenylalanine and tyrosine being at least equal to 0.002 on the same respective molar basis, the respective molar proportions of isoleucine, leucine, valine, tryptophan, phenylalanine, and tyrosine being selected from the above molar ranges thereof so that the ratio of the combined molar proportions of isoleucine, leucine, and valine to (a) the molar proportion of tryptophan is within the numerical range from 40 to 300, and to (b) the combined molar proportion of phenylalanine and tyrosine is within the numerical range from 15 to 135.

In 1949, Dr. William C. Rose published studies establishing minimum amounts of 20 amino acids needed by healthy persons--the "Rose pattern." Of these, eight amino acids are identified as essential,2 two are sometimes identified as semi-essential,3 and the balance are deemed non-essential.4 The principal function of the normal liver in the human body is to regulate the levels of certain amino acids in the blood.

Proteins are normally obtained from food and are broken down into amino acids before they can be used for nourishment in the human body. Due to illness, surgery, or injury, however, some persons cannot gain nutritionally adequate quantities of protein through the alimentary tract and quickly develop malnutrition. Intravenous administration of amino acids formerly compensated partially for a patient's protein deficit. In the 1940's amino acid hydrolysates were prepared from proteins such as milk protein or hen's egg protein. By the mid-1960's, however, all of the amino acids necessary for human nutrition had become available in pure form at a reasonable cost. Further, surgical techniques for administration of nutritionally adequate quantities of amino acids had also become available.

The technique of providing nutritionally adequate quantities of amino acids intravenuously--hyperalimentation5--could not be used with all patients, however. Most people with liver disease can be treated by controlling their diet. Some patients with liver disease, particularly cirrhosis, develop hepatic encephalopathy or coma when fed protein. Consequently, malnutrition presents a serious problem to these patients.

Ammonia is formed in the gut from the breakdown of proteins. Based on observations showing an abnormally high ratio of ammoniagenic amino acids in the plasma of patients with certain liver diseases, it is thought by some that when the liver is not functioning properly excess ammonia builds up resulting in hepatic encephalopathy. In the early 1970's ammonia was thought to be the principal cause of hepatic encephalopathy and this theory is still widely accepted.

In a 1973 paper,6 Dr. Rudman classified the ammoniagenic tendencies of amino acids metabolized in patients with liver disease.7 Conventional treatments for controlling plasma ammonia levels, however, were not always successful; they sometimes caused malnutrition and did not always prevent hepatic encephalopathy.

In 1971, Dr. Fischer published an article8 presenting an alternative hypothesis that some of the neurological and cardiovascular complications of hepatic failure could be caused by abnormally high levels of "aromatic" amino acids9 relative to "branched chain" amino acids10 in the plasma of liver-diseased patients. Fischer suggested that the aromatic and branched chain amino acids compete for passage into the brain, across the "blood-brain barrier." Branched chain amino acids are not regulated by the liver but, rather, through other metabolic pathways independent of the liver. The levels of aromatic amino acids in the plasma are regulated by the liver and, consequently, there is some relationship between a damaged liver and abnormally high levels of aromatic amino acids in the plasma. In liver-diseased patients the levels of aromatic amino acids are elevated, while the levels of branched chain amino acids are decreased. Fischer thought that aromatic amino acids were precursors of false neurotransmitters which disrupted normal brain functioning in patients with hepatic encephalopathy. Thus, Fischer proposed that hepatic encephalopathy may be caused by the excessive levels of aromatic, relative to branched chain, amino acids in the plasma.

It was known that certain amino acids were elevated in patients with specific types of liver disease. Prior to 1974, "normalization" of amino acid levels for patients with a specific enzyme deficiency was known. In normalization, a diet abnormally low in certain amino acids is given to patients whose levels in those amino acids prior to treatment were abnormally high, and vice versa. Normalization was not, however, a conventional treatment for patients with liver disease.

In 1972, Fischer entered into a collaborative research effort with American to formulate an amino acid product for patients with liver disease. The work initially involved two intravenous amino acid formulations: FreAmine, an amino acid product for patients with kidney disease, and FreAmine E, an experimental formulation. The results of that research effort, up to early 1973, were published in 1974.11

In the 1974 article Fischer suggested modifying amino acid formulations to normalize the pattern of amino acids in the plasma of patients with liver disease. The FreAmine E preparation tested by Fischer consisted of only the eight essential amino acids conforming to the Rose pattern. To Fischer's surprise, FreAmine E failed to increase the low plasma concentration of branched chain essential amino acids even though FreAmine E contained 2 1/2 times the minimum requirements of branched chain amino acids recommended by Rose.

Following oral presentation in May 1973 of the report later published as the 1974 Fischer article, Fischer and Dr. Yoshimura devised a modified formulation intended to normalize plasma amino acid levels in liver-diseased patients. They called the formulation "F080." The ratio of branched chain to aromatic amino acids was much higher in the F080 formulation than in FreAmine. Fischer performed animal tests with the F080 formulation in 1974. Animals in a coma or near comatose states with simulated liver disease awakened when given the F080 formulation.

On August 17, 1972, Ghadimi filed an application for an "Injectable Amino Acid Composition Commensurate to the Anabolic Need of the Body and Method of Using Same." U.S. patent No. 3,832,465 (the Ghadimi patent) issued to Ghadimi on August 27, 1974, claiming a nutritional amino acid formulation. The Ghadimi patent does not address the specific problems of patients with liver disease but, rather, provides a formulation for the nutritional needs of patients from a variety of age groups--newborn to adult.

On February 3, 1975, Fischer filed application serial No. 546,689. On the basis of that application, the Fischer '529 patent issued on April 13, 1976. The claims were allowed without discussion or amendment.

In May 1979, American began selling an amino acid formulation for liver-diseased patients made in accordance with claim 1 of the Fischer '529 patent as an enteral (food) product under the trademark "Hepatic-Aid." Travenol introduced a competing enteral amino acid formulation for nutritional support for liver-diseased patients, in the United States market in September 1981, under the trademark "Travasorb Hepatic."

On July 12, 1982, American filed with the U.S. International Trade Commission a complaint alleging violations by Travenol of 19 U.S.C. Sec. 1337 (1982). The Commission investigated American's claim of infringement of claims 1, 5, and 14, and of contributory infringement and inducing infringement of claims 1, 5, 6, 7, 9, and 14, of the Fischer '529 patent. In an initial determination issued May 23, 1983, the presiding official held that the Fischer '529 patent is not invalid, is enforceable, and is not infringed by Travenol.

The presiding official rejected Travenol's contentions that the claims are invalid under 35 U.S.C. Sec. 112 for vagueness, ambiguity, or indefiniteness, under section 102(e) in view of the Ghadimi patent, or under section 103 for obviousness in view of the 1974 Fischer article and the Ghadimi patent. While American was found to have misrepresented, and to have failed to disclose, certain information to the Patent and Trademark Office (PTO) during prosecution, that information was not found to be material, and the Fischer '529 patent was held to be enforceable. The Fischer '529 patent was not found to be infringed by Travenol because of differences between the claimed and the Travenol formulations. The presiding official evaluated equivalence at the time the invention was made and declined to recognize the reduced amounts of one amino acid in the Travenol formulation as equivalent to the claimed amount of that amino acid despite evidence that the minimum requirement of that amino acid was reduced subsequent to the time of Fischer's invention. In view of the alternative ammoniagenic theory of causation of hepatic encephalopathy, the presiding official concluded that American failed to meet its burden of proving that Travasorb Hepatic worked in substantially the same way as the claimed formulation.

The Commission declined to review the initial determination of no violation and, accordingly, the initial determination became final on June 22, 1983.12 American appeals.

Three principal questions are raised by the parties13 on appeal:

(1) whether the asserted claims are invalid under either section 102(e) or section 103;

(2) whether the appealed claims are unenforceable because of alleged misrepresentations or omissions made by American during prosecution of the Fischer '529 patent; and

(3) whether Travasorb Hepatic infringes the asserted claims under the doctrine of equivalents.

Travenol contends that, while the Commission's determination that Travenol has not violated section 1337 is correct, the asserted claims are clearly invalid under section 102 or section 103 or both.

Travenol asserts that the Commission entered a variety of clearly erroneous findings of fact with respect to the teachings of the references. Many of these, however, are straw men stuffed with Travenol's misinterpretations of the Commission's findings. Further, we are not persuaded that the biochemistry involved was "predictable," as Travenol claims. Travenol ignores the findings and the evidence with respect to the complexity of normalizing plasma amino acid levels and with respect to the uncertainty surrounding the competing theories of treatment of hepatic encephalopathy.

The Fischer '529 patent claims a specific range of amino acid formulations for patients with liver disease. Contrary to the implication of some of Travenol's arguments, it does not claim the concept of normalizing plasma amino acid levels in the abstract.

Travenol asserts that the claims in issue are anticipated by the Ghadimi patent. In Kalman v. Kimberly-Clark Corp.,14 this court stated:

A party asserting that a patent claim is anticipated under 35 U.S.C. Sec. 102 must demonstrate, among other things, identity of invention. * * * [I]dentity of invention is a question of fact, * * * and one who seeks such a finding must show that each element of the claim in issue is found, either expressly described or under principles of inherency, in a single prior art reference, or that the claimed invention was previously known or embodied in a single prior art device or practice. * * *

Travenol argues that the Commission's failure to recognize Ghadimi's teachings with respect to normalization is clearly erroneous and that the Commission erred in emphasizing the nonoverlapping portions of the ranges in the Ghadimi patent. We disagree.

First, we do not review the findings of the Commission under the clearly erroneous standard applicable to the findings of a district court as provided in Fed.R.Civ.P. 52(a). We review the factual findings of the Commission to determine whether they are unsupported by substantial evidence. We are not bound by the Commission's legal conclusions.15

While the Ghadimi patent teaches normalization of plasma amino acid levels, it does not disclose a formulation falling within the ranges of amino acids claimed by Fischer. Travenol's attempt to make out anticipation by drawing the court's attention to the overlap between portions of the Ghadimi ranges and the claimed ranges of amino acids--16 of 19 amino acids--appears to be little more than a diversion of attention from the three claimed amino acid ranges which Ghadimi does not disclose.16

While Ghadimi discloses a formulation using all of the 19 amino acids in the Fischer patent, it does not disclose any amino acid formulation that meets the asserted claims. There is no identity of invention. Thus, we conclude that the Commission's finding, that the asserted claims of the Fischer patent are not anticipated by Ghadimi, is supported by substantial evidence. The asserted claims of the Fischer '529 patent are not invalid under section 102(e).

Travenol also argues that the asserted claims are invalid for obviousness under section 103 in view of either the 1974 Fischer article or the Ghadimi patent or both. Several of the Commission's findings of fact and statements of the law are designated by Travenol as clearly erroneous17 and erroneous, respectively. The Commission employed the Graham v. John Deere Co.18 criteria in determining whether the asserted claims would have been obvious under section 103. Travenol's attack is leveled primarily at the Commission's findings with respect to the unpredictability of normalization techniques and with respect to the teachings of the 1974 Fischer article and the Ghadimi patent in terms of what would have been obvious to one of ordinary skill in the art.

There is no dispute with respect to the scope and content of the prior art--all parties agree that the 1974 Fischer article and the Ghadimi patent were properly considered the most pertinent prior art.

As Travenol correctly argues, the 1974 Fischer article plainly suggests that "the approach to total parenteral nutrition in patients with hepatic disease should include the normalization of plasma amino acid patterns." In view of the abnormal plasma amino acid pattern arising in liver disease, Fischer states that normalization of such patterns would be "beneficial." Ghadimi suggests the monitoring and adjustment of the plasma amino acid levels in formulating amino acid preparations for intravenous administration. Ghadimi also discloses that

in the present state of knowledge it is not known exactly what level of free amino acids in the blood should be considered toxic. The magnitude of the problem becomes apparent when it is realized that "toxic effect" means long-range effects on the growing brain and hence subnormal brain function or even gross mental retardation not amenable to therapy by the time the long-range effect is detectable by conventional techniques.

The "toxic" effects addressed by Ghadimi, however, are not the same problems to which the Fischer '529 patent is directed.

While Ghadimi observed that "infusion of branched-chain amino acids (leucine, isoleucine, valine) substantially in excess of minimal or even 'safe' requirements (double the minimum) did not result in unduly raised amino acids in the plasma," there is no recognition in the Ghadimi patent that a formulation with a high ratio of branched chain to aromatic amino acids, such as that claimed by Fischer, would be beneficial to patients with severe liver disease. The failure of that infusion to raise plasma amino acid levels a certain degree suggested to Ghadimi only that the metabolic pathways of these amino acids lie outside the liver.

The Ghadimi patent is directed to providing adequate nutritional support to patients of all ages, without regard to liver disease. The claimed Fischer formulation is not disclosed by Ghadimi and is directed to a different class of users with specific unique nutritional problems. The 1974 Fischer article discloses no specific formulation. It suggests only that, in general, normalization may be beneficial. It does not, however, direct one how to proceed in developing an amino acid formulation that would result in normalized plasma amino acid levels while avoiding the risk of hepatic encephalopathy. In view of the mystery enshrouding hepatic encephalopathy, the Commission found that it was not clear that the concept would work. The Commission recognized that "obvious to try" is not a legitimate test of patentability under section 103.19

The level of ordinary skill in the art was found to be high, a finding repeatedly emphasized by Travenol in its brief on appeal. Travenol disputes that one possessing such a high level of skill would have found the technique of normalization unpredictable. The record clearly establishes that the prevailing wisdom at the time of the invention was that patients with hepatic encephalopathy were not fed protein for any reason, because it made their conditions worse. These patients were thought to be intolerant of protein. American's expert witness testified that, given the lack of background information on the causes of liver disease, normalization was unpredictable at the time of Fischer's invention. Thus, the Commission's finding is supported by substantial evidence.

Travenol also asserts that American failed to show any new and surprising result and that the Commission erred in accepting the alleged criticality of the claimed ranges. The burden of proving invalidity, however, was on Travenol. American is under no compulsion either to prove a new and surprising result or to prove the criticality of the claimed ranges of amino acids. Rather, the burden was on Travenol to establish the lack of new and surprising results or the lack of criticality. Travenol failed to prove either.

We hold that the Commission did not err in concluding that the Fischer '529 patent is not invalid for obviousness under section 103.

Travenol contends that American committed inequitable conduct by virtue of American's alleged breach of the duty of candor during prosecution of the Fischer '529 patent. The Commission found that, while at least one misrepresentation and a number of omissions were made during prosecution, none of these was material. Thus, no inequitable conduct was found. The legal standard of materiality applied by the Commission is cited by Travenol as error.

We agree that the Commission did not apply the standard of this court as set out in Driscoll v. Cebalo.20 Because of the discussion of the issue of enforcement there, and because of our disposition of this appeal on the basis of noninfringement, we see no need to examine the issue of enforceability de novo.

American did not allege literal infringement but, rather, alleged that Travenol infringes the asserted claims under the doctrine of equivalents. The Commission, however, declined to find infringement. The same result was found to be achieved by both the claimed product and Travasorb Hepatic--providing adequate nutrition to patients with severe liver disease who cannot tolerate normal proteins without suffering hepatic encephalopathy. However, Travenol was not found to use substantially the same means to achieve that result.

While Travenol copied the high ratio of branched chain amino acids to aromatic amino acids from the Fischer '529 patent, the molar ranges of five amino acids in the Travenol formulation fall below the claimed ranges.21 The Fischer '529 patent teaches that the ratios of all 19 amino acids to each other should be maintained. Subsequently, as knowledge in the art advanced, some of Fischer's claimed ranges were found to be too stringent. The Commission concluded, however, that the asserted claims could not be expanded by the doctrine of equivalents "as new information becomes available, and it is learned which ingredients and ratios in the formulation are important and which are not." The Commission found further that the proportions of many of the amino acids were changed in accordance with the competing ammoniagenic theory of treatment of hepatic encephalopathy.

American correctly contends that the Commission erred in determining equivalence at the time of invention without regard to subsequent developments in the art. In American's view the levels of several amino acids of Travasorb Hepatic falling below the claimed ranges would not impair the function of the alleged infringing product. American also argues that the Commission erred in restricting the claims in view of the prior art and in comparing Travasorb Hepatic to American's Hepatic-Aid, rather than to claim 1. American urges that the evidence compels a finding of infringement.

Travenol, however, argues that the claims are narrow and that the prior art is more pertinent to the claimed invention than is Travasorb Hepatic. Relying on Dr. Fischer's testimony regarding risk to the patient of formulations outside of the claimed ranges, Travenol asserts that there is no equivalence.

The Commission defends its finding of no infringement by accurately focusing on the difference in means between the claimed and the alleged infringing formulations. Citing the uncertain state of current knowledge of the mechanisms of hepatic encephalopathy, the Commission urges that American has failed to meet its burden of proof on infringement.

A finding of infringement under the doctrine of equivalents is a finding of fact.22 Accordingly, we must affirm the Commission's finding unless it is unsupported by substantial evidence.23

The Commission recognized that the critical questions in this case with regard to infringement were "whether substantially the same ingredients achieve substantially the same results in substantially the same way, whether those with ordinary skill in the art would recognize certain ingredients as the equivalents of one another, and whether certain ingredients are not an essential part of the claimed invention." Contrary to American's assertion, the Commission assessed infringement with respect to the asserted claims, not American's commercial product.

Using a "common factor" approach to derive a comparable sample of Travasorb Hepatic to test against claim 1, the Commission found that Travasorb Hepatic fell outside of the claimed amino acid ranges with respect to 5 of 19 amino acids.24 Of these, four--serine, alanine, proline, and glycine--are classified by Rose as nonessential. Thus, the lower-than-claimed levels of these nonessential amino acids would not impair the nutritional value of the formulation. Threonine, however, is classified by Rose as essential and, for that reason, its concentration in Travasorb Hepatic could have been considered to impair the nutritional value of the formulation at the time of the invention.

Subsequently, however, research revealed that the minimum requirement of threonine was much lower than the range claimed by Fischer. The Commission recognized that this was not known at the time of the intention and held that the scope of the claims of the Fischer patent cannot be read to cover subsequent advances. That conclusion is contrary to this court's precedent.

In Hughes Aircraft Co. v. United States,25 this court, as had the Court of Claims,26 held that variations in the invention, made possible by subsequent advances in the art, do not allow the accused infringing device to escape the "web of infringement." An appropriate range of equivalents may extend to post-invention advances in the art in an appropriate case.

As the Commission noted, the specification of the Fischer '529 patent teaches (1) that the relationships among amino acids set forth in the chart in claim 1 is important and (2) that the ratio of branched chain amino acids to aromatic amino acids is important. Travasorb Hepatic falls literally within the claimed range of ratios of branched chain amino acids to aromatic amino acids. The specification expressly teaches that, while the inclusion of nonessential amino acids is desirable, the formulation may include only essential amino acids. The level of threonine in the claimed formulation is maintained in a specific range not for reasons associated with the ratio of branched chain amino acids to aromatic amino acids but, rather, for nutritional reasons. Thus, subsequent advances in the prior art, indicating that less threonine is needed for nutritional purposes than previously had been thought, are consistent with the specification. The Commission erred in holding that subsequent advances could not be included within the scope of protection afforded the asserted claims. That error, however, does not merit reversal.

In view of our disposition of the remaining question--whether Travasorb Hepatic works in substantially the same way as the claimed invention--we need not, and do not, decide whether the Commission's finding, that Travasorb Hepatic and the claimed formulation do not perform substantially the same function, is unsupported by substantial evidence.

The difference in threonine levels is significant for another reason. Threonine is identified by Dr. Rudman in his 1973 article as among the most ammoniagenic amino acids. The Commission found the theory, that ammonia causes hepatic encephalopathy, has continuing vitality and found that many of the amino acids whose proportions in Travasorb Hepatic were changed by Travenol are important in their effect on ammonia. Fischer testified that he thought the ratio of branched chain amino acids to aromatic amino acids was critical to his invention and was the key to hepatic encephalopathy. Fischer recognized however that there was no certainty with respect to knowledge about the causes of hepatic encephalopathy. The record clearly establishes that the cause and mechanism of hepatic encephalopathy were, and remain, unknown.

In view of that uncertainty, the Commission could readily have concluded that American failed to meet its burden of proving infringement. The record establishes the continued vitality of the competing theory of treating hepatic encephalopathy--reducing the levels of ammoniagenic amino acids. While we do not adopt the Commission's reasoning, we conclude that the Commission's finding, that Travasorb Hepatic does not infringe the asserted claims of the Fischer '529 patent under the doctrine of equivalents, is supported by substantial evidence.

In summary, the asserted claims of the Fischer '529 patent are not anticipated by the Ghadimi patent, nor are they invalid for obviousness under section 103 in view of the 1974 Fischer article or the Ghadimi patent, either alone or in combination. We conclude that the Commission's finding that the asserted claims of the Fischer '529 patent are not infringed under the doctrine of equivalents is supported by substantial evidence. In view of the competing "ammoniagenic" theory of hepatic encephalopathy, American did not meet its burden of proof in demonstrating that Travasorb Hepatic functions in substantially the same way as the claimed formulation. Accordingly, we sustain the Commission's determination that Travenol has not violated 19 U.S.C. Sec. 1337.

AFFIRMED.