A Silicone Rubber Implant to Supplement the Keller Toe Arthroplasty

Alfred B. Swanson, M.D. Sydney L. Biddulph, M.D. Carl-Göran Hagert, M.D.

Arthroplasty of the first metatarsophalangeal joint, as described by Keller in 190411, was one of the first in a long list of operative procedures proposed for the solution of the many problems centered around the great toe. Although it has never been accepted universally, the Keller arthroplasty, because of its simplicity and efficiency, has remained with us for almost three-quarters of a century. This paper describes the use of a silicone rubber implant which has been designed to supplement the Keller procedure and overcome more of its disadvantages.

Historical Background

The Keller procedure gave the surgeon an opportunity to tackle the problems of the great toe by resection arthroplasty. However, it had certain inherent disadvantages. These disadvantages led surgeons to either modify the operation or look for alternative procedures. At present these approaches can be classified into two groups:

  1. Those directed at the deformity itself, and
  2. Those directed at correcting the underlying etiological factors.

Procedures focusing upon the deformity itself are arthroplasties, aimed at maintaining motion1,6,11,13,19,27, and arthrodesis of the first metatarsophalangeal joint7,16,18,28,29,30. Procedures aimed at correcting the underlying associated anatomical features are soft tissue operations10,14,15 and metatarsal osteotomies9,17,31.

Probably the Keller arthroplasty is still the most widely practiced procedure, but disappointing results have been reported2,18. It is easy to misjudge the amount of bone to be removed. If too little is removed, a stiff, painful pseudar-throsis may result. On the other hand, excessive bone removal results in a short, floppy toe, with resulting insecurity and instability; virtually a physiological amputation. Young people complain of the lack of power at take-off, and young females do not care for the cosmetic effect. The wide joint space gradually narrows and results become progressively worse with the passing of time2. Also with excessive shortening of the great toe the second toe is uncovered and becomes subject to pressure symptoms. Painful corns may develop. Results following the treatment of Hallux rigidus are even more disappointing2.

Thomas27, in an effort to maintain "joint space" following arthroplasty, prevented collapse of the proximal phalanx by splinting with Kirschner wires. Improved results for up to one year of follow-up were reported. It is unlikely, however, that progressive shortening would be permanently prevented by this procedure.

Few surgeons perform the Mayo procedure as too much removal of bone may cause serious alterations in the weight-bearing characteristics of the foot.

Arthrodesis of the metatarsophalangeal joint provides a stronger foot in the younger patient. Fusion rates, however, range between 70 and 80 per cent7,16,18,28,29,30. Abnormal strain is often concentrated on the interphalangeal joint of the great toe, especially if dorsiflexion is limited. Later, degenerative changes may ensue. Fusion in an incorrect position is a frequent problem. Too much varus, plantar or dorsiflexion causes excessive pressure, and too much valgus crowds the lateral toes, or may produce overlapping.

Soft tissue procedures, more successfully used in the younger patient, rarely succeed in the long-standing case. This is sequential to the fact that no account is taken of secondary patho-anatomical changes (lateral capsular contraction, cartilage denudation).

On the average, metatarsal osteotomy produces results far below expectations. The reasons for failure were cogently recorded by Bonney and MacNab. The aim of the operation, namely, to preserve the function of the metatarsophalangeal joint, is often not realized. This may be because the segment of metatarsal head brought into contact with the proximal phalanx may be denuded of cartilage; or it may be due to:

  1. Failure to correct soft tissue contracture.
  2. Failure to maintain metatarsal correction.

A stiff, painful, metatarsophalangeal joint is the wage of failure. Elevation of the first metatarsal head causes undue weight to be borne by the second metatarsal with subsequent metatarsalgia2,9,17,31. Nonunion and avascular necrosis of the head of the first metatarsal may occur.

In an attempt to overcome the problems encountered with the Keller arthroplasty, the senior author designed a silicone-rubber implant to prevent the short, floppy toe and the painful pseudarthrosis.

The idea was first conceived in 1952, at a time when the senior author was particularly interested in foot problems. A metal prosthesis consisting of an intramedullary stem fitted to a hemispherical cap was designed (Fig. 1 -A). The stem fitted into the medullary canal of the first metatarsal. The cap fitted snugly over the end of the bone and substituted for the metatarsal head. A relatively small number of cases was done with unexciting results. As with many metal prostheses, problems were encountered with absorption of bone.

With the introduction of silicone rubber as an implant material, new life was given to the concept of augmenting the toe arthroplasty with a prosthesis. Five years ago a silicone-rubber implant was developed to replace the first metatarsal head (Fig. 1 -B). It had some of the disadvantages of the previous metal implant. Stability of the prosthesis in the first metatarsal was uncertain. The implant was fabricated from heat-vulcanized, medical-grade, silicone elastomer stock (Silastic brand) manufactured by Dow-Corning Corporation, Midland, Mich.

Experience gained in designing implants of silicone rubber for end-bearing amputation stumps, overgrowth problems in juvenile amputees, and for arthroplasties in arthritic and destroyed joints (MP joints, PIP joints, trapezium, lunate, scaphoid, wrist joint, ulnar head, radial head, elbow joint, shoulder joint, and knee joint) has been used in the development of the toe implant20-26.

Because of the instability experienced with the earlier implant, a decision to design an implant for the nonweight-bearing side of the joint was reached. The base of the proximal phalanx rather than the head of the first metatarsal is replaced (Fig. 1 -C).

The stem of the implant fits into the medullary canal of the proximal phalanx with the cap replacing the proximal third and base of the phalanx. In situ the cap articulates with the head of the metatarsal. Joint stability depends on the surrounding capsule and ligaments.

The head of the metatarsal can be trimmed without the need to replace its weight-bearing function by an implant which would be unstable because of the excessive weight-bearing thrust developed in walking. The implant placed at the phalangeal aspect of the joint preserves motion, is not subjected to the forces found at the metatarsal head, and is very stable. It maintains its length and provides a smooth articulating surface.

In vitro testing of the implant has been done on the test machine. After 160 million repetitions with an excursion of 90 deg., no evidence of wear has been found. In terms of human effort and day-to-day living, this would be the equivalent to walking a distance of 160,000 miles.

Five sizes designed to cope with most variations in bone size have been produced (Fig. 2 ).

Indications for Use of the Silicone Insert

  1. Hallux valgus.
  2. Hallux rigidus.
  3. Combined with the Hoffmann or Clayton procedures for rheumatoid arthritis affecting the feet.

Preparation of the Implant

Medical-grade silicone elastomers are extremely inert; however, if they become contaminated by lint, fingerprints, or other foreign materials they can cause a foreign-body reaction. It is therefore important that the implant be thoroughly cleaned and sterilized. It should be washed thoroughly in a hot-water soap solution to remove possible surface contaminants. Synthetic detergents or oil-based soaps should not be used as these soaps may be absorbed and subsequently leach out to cause a tissue reaction. The implants must be rinsed copiously in hot water followed by thorough rinsing in distilled water. The material is heat-resistant and can be autoclaved. Every effort should be made to eliminate the chances of contamination with other foreign bodies throughout the entire procedure. The implants should be handled with instruments rather than with gloves.


Using a dorsomedial approach, a longitudinal skin incision, slightly curved, is made along the medial aspect of the metatarsophalangeal joint. Care is taken to avoid injury to the small dorsal sensory nerves in the area. The fascia and the medial capsule are dissected, making a flap distally based on the proximal phalanx. This flap is reflected for later resuture. If a bursa is present it is excised and the metatarsophalangeal joint opened. The proximal third of the basal phalanx is resected with drill holes and an osteotome, or with an air drill or motor saw (Fig. 3 ). The exostoses of the metatarsal head are removed on both lateral and plantar surfaces so that a smooth rounded metatarsal head is presented. The sesamoid bones are rarely excised. Using a drill, broach or curette, the canal of the proximal phalanx is shaped to accept the implant stem.

The implant stem should fit snugly into the canal and the collar of the implant should fit firmly against the proximal cut surface of the phalanx. This type of fit will require a proper implant choice from the five available sizes. The stem of the prosthesis may be cut to appropriate length. No trimming should be done elsewhere on the implant. The toe should then be easily brought over into the corrected position.

Further release of the lateral capsule, adductor muscle attachments, or rerouting or lengthening of the long extensor may be indicated. There should be a good and unrestricted range of motion of the toe before closure. With the toe in the corrected position, the fascial-capsular flap is firmly sutured to the bone through small drill holes (Fig. 4 ).

A 4-0 dacron suture using a buried-knot technique holds the capsule to the medial aspect of the metatarsal shaft in the corrected position. The skin is closed and a small drain is placed in the wound. The toe is held in a bunion-type splint using a voluminous pressure dressing. The splint is kept on the toe for approximately five days. A dynamic splint (Fig. 5 ) is then applied and used continuously for approximately one month. It is then worn as a night splint for four to six weeks. Early flexion-extension movements of the toe are encouraged. Guarded weight-bearing is allowed in three weeks.

The toe implant, Keller-type procedure can also be used for the great toe when the heads of the other metatarsals are resected through a transverse incision in the typical procedures of Hoffmann or Clayton3,8. The transverse incision is carried slightly further distally at its medial end and the above-outlined procedure for the great toe can easily be carried out.

Figs. 6 and 7 , Figs. 8 and 9


The operation described has been performed on 35 patients during the last three years.

There has been no incidence of infection, or of fracture or dislocation of the implant. Pain relief, especially in patients with hallux rigidus, has been most satisfying. The length of the great toe was maintained, resulting in improved power. The floppy great toe has been obviated and a much improved cosmetic result has been achieved.

Hallux valgus recurred in one patient, but without any symptoms. This complication should be preventable by meticulous capsular repair.

In rheumatoid arthritis, collapse of the great toe following the classic Hoffmann procedure is often seen (Fig. 10 ). This collapse may result in a painful pseudarthrosis. This complication did not occur when the arthroplasty procedure was supplemented with a great-toe silicone-rubber implant.


The fact that many surgical procedures have been developed for the treatment of hallux valgus indicates the complexity of the problem.

Uncertain surgical results have led to a continuing search for a consistently successful operation. The Keller operation has been a satisfactory procedure in many cases. Its poor results were due to two factors:

1. The first metatarsal constitutes the major part of the anterior portion of the longitudinal arch. It carries a large proportion of the weight exerted upon the forefoot. Interference results in a shift of this weight laterally onto the second metatarsal, with the development of pain and callosities. The great toe ordinarily provides grip, especially on rough terrain, even in the shod foot. A short, floppy, flail toe results in insecurity and instability. The great toe acts as the major springboard in forward propulsion. It is the last part of the foot to leave the ground and in so doing provides vital force and lift in transferring body weight from the hind foot to the forefoot. Lack of this power interferes with a normal gait pattern.

2. Recurrence of symptoms. Usually this problem is the result of not removing enough bone from the proximal phalanx. The joint space narrows and a painful, stiff pseudarthrosis ensues.

The silicone rubber prosthesis provides a smooth articulating surface for the resected joint. Used as a supplement to the Keller-type arthroplasty, loss of length of the great toe is prevented. This advantage provides greater strength and stability to the forefoot.

In combination with the Hoffmann or Clayton procedures the implant provides a smooth surface to articulate with the remodeled head of the first metatarsal, even when all articular cartilage has been removed. The painful pressure symptoms are relieved yet strength and weight-bearing characteristics are improved. The Mayo type of resection may also be improved in this fashion.


A new toe implant of heat-molded silicone rubber to supplement the Keller and Hoffmann arthroplasty is introduced. The technique is described briefly.

Supported by a grant from the John A. Hartford Foundation.

Alfred Swanson is Chief of the Orthopaedic Department, Blodgett Memorial Hospital, and Chief of the Training Program and Orthopaedic Research. Sydney Biddulph is a fellow in Orthopaedic Research, Johannesburg, South Africa. Carl-Göran Hagert is a fellow in Orthopaedic Research, Johannesburg, South Africa.

1. Bingham, Robert, The stone operation for hallux valgus, Clinical Orthopaedics, 17:366, 1960. 
2. Bonney, G., L. W. and I. MacNab, Hallux valgus and hallux rigidus, a critical survey of operative results, J. Bone and Joint Surg., 34-B:366, 1952. 
3. Clayton, M. L., Surgery of the forefoot in RL arthritis, J. Bone and Joint Surg., 42-A:523, 1960. 
4. Cleveland, M., and E. M. Winant, An end-result study of the Keller operation, J. Bone and Joint Surg., 32-A: 163, 1950. 
5. Fowler, G. R., A treatise on surgery, Vol. II, W. B. Saunders, Philadelphia, 1907, page 622. 
6. Girdlestone, G. R., and H. J. Spooner, A new operation for hallux valgus and hallux rigidus, J. Bone and Joint Surg., 19:30, 1937. 
7. Harrison, M. H. M., and F. J. Harvey, Arthrodesis of the first metatarsophalangeal hallux valgus and rigidus, J. Bone and Joint Surg., 45-A:47l, 1963. 
8. Hoffmann, P., An operation for severe grades of contracted or claw toes, Am. J. Orth. Surg., 9:441, 1911. 
9. Hohmann, G., Fuss und Bein, J. F. Bergmann, Munich, 1948, p. 169. 
10. Joplin, R. J., Sling procedure for correction of splay foot, metatarsus primus varus, and hallux valgus, J. Bone and Joint Surg., 32-A:779, 1950. 
11. Keller, W. L., Surgical treatment of bunions and hallux valgus, New York Med. J., 80:741, 1904. 
12. Lapidus, P. W., Operative correction of metatarsus varus primus in hallux valgus, Surg. Gyn. and Obst., 58:183,1934. 
13. Mayo, C. M., The surgical treatment of bunions, Annals of Surg., 48:300, 1908. 
14. McBride, E. D., A conservative operation for bunions, J. Bone and Joint Surg., 10:735, 1928. 
15. McBride, E. D., The McBride bunion hallux valgus operation, J. Bone and Joint Surg., 49-A: 1675, 1967. 
16. McKeever, D. C, Arthrodesis of the first metatarso-phala?geal joint for hallux valgus and hallux rigidus and metatarsus primus varus, J. Bone and Joint Surg., 34-A:129, 1952. 
17. Mitchell, C. L., et al., Osteotomy-Bunionectomy for hallux valgus, J. Bone and Joint Surg., 40-A:41-58, 1958. 
18. Moynihan, F. J., Arthrodesis of the MP joint of the great toe, J. Bone and Joint Surg.,49-B:544, 1967. 
19. Silver, D., The operative treatment of hallux valgus, J. Bone and Joint Surg., 5:225, 1923. 
20. Swanson, A. B., A flexible implant for replacement of arthritic or destroyed joints in the hand, Inter-Clin. Information Bull., 6:16-19, 1966. 
21. Swanson, A. B.,' Bone overgrowth in the juvenile amputee and its control by the use of silicone rubber implants, Inter-Clin. Information Bull., 8:9-18, 1969. 
22. Swanson, A. B., Finger joint replacement by silicone rubber implants and the concept of implant fixation by encapsulation. International Workshop on Artificial Finger Joints, Suppl. to Ann. Rheum. Dis., 28:47-55, Sept. 1969. 
23. Swanson, A. B., Improving the end-bearing characteristics of lower-extremity amputation stumps, Inter-Clin. Information Bull., 5:1-7, 1966. 
24. Swanson, A. B., Silicone rubber implants, replacement of arthritic or destroyed joints of the hand, Surg. Clin, of N. A., 48:1113, 1968. 
25. Swanson, A. B., The need for early treatment of the rheumatoid hand, J. Mich. State Med. Soc, 60:348, 1961. 
26. Swanson, A. B., and Y. Yamuachi, Silicone rubber implants for replacement or arthritic or destroyed joints, Scientific Exhibit, American Academy of Orthopaedic Surgeons, Chicago, January 1968. 
27. Thomas, F. B., Keller's arthroplasty modified. A technique to ensure post-operative distraction of the toe, J. Bone and Joint Surg., 44-B:356, 1962. 
28. Thompson, F. R., and R. T. McElvenny, Arthrodesis of the first MP joint, J. Bone and Joint Surg., 22:555, 1940. 
29. Wilson, C. L., A method of fusion of the MP joint of the great toe, J. Bone and Joint Surg.,40-A:384, 1958. 
30. Wilson, J. N., Cone arthrodesis of the first MP joint, J. Bone and Joint Surg., 49-B:98, 1967. 
31. Wilson, J. N., Oblique displacement osteotomy for hallux valgus, J. Bone-and Joint Surg.,45-B:552, 1963.