Dr. Anthony A. Schepsis

Coastal Orthopedics
Beverly, MA
Professor of Orthopedic Surgery
Boston University School of Medicine

Connect With Dr. Schepsis




Reconstruction of the Medial Patellofemoral Ligament

 Technical Update: Reconstruction of the Medial Patellofemoral Ligament for Recurrent Patellar Instability 

Anthony A. Schepsis, M.D., Jack Farr, M.D. 

Recurrent lateral patellar instability after traumatic patellar dislocation or subluxation is a commonly encountered problem. In recent years, the importance of the medial patellofemoral ligament (MPFL) as the primary soft tissue restraint to lateral translation of the patella has been corroborated by several authors (3, 4, 7, 11, 13). As an extension of these investigations, the MPFL has been likewise, established as the primary passive soft tissue restraint to pathologic lateral translation of the patella. Thus, by definition, the MPFL is always injured to some extent during traumatic lateral patellar dislocations [4, 11, 7]. It then follows that reconstruction of the MPFL is often recommended for patholaxity of the medial patellar stabilizers with the understanding that other medial tissues contribute (e.g., medial patellotibial), but are much less biomechanically important. Although historically numerous procedures have been described for addressing insufficiency of the medial restraints, (e.g. reefing, advancements, and nonanatomic tendon transfer procedures) most were designed without the current knowledge base of the MPFL and associated anatomic medial restraints. As a result, some of these earlier procedures have, at times, created the potential for abnormal forces and contact areas (leading to elevated stress) of the patellofemoral articulation. The current goal of patellofemoral surgery is correct pathology with inflicting iatrogenic pathology. That is, in this case, to recreate the restraint of the MPFL without creating abnormal biomechanics. As such, reconstruction is not an over-constraining procedure and does not seek to address static patellar position, nor is it indicated for primary problems of pain and/or arthritis. 

The following technique describes an approach to the reconstruction of the MPFL utilizing a free tendon graft. Proper selection of patellar and femoral attachment sites, anatomometric testing (as opposed to isometric), proper length establishment and tensioning, and fixation will be described. The graft is secured to the femoral attachment site with an interference fit bioabsorbable screw. Two variations for patellar attachment of the graft, with either suture anchor fixation or interference fit will be detailed. This procedure may be performed in association with tibial tubercle realignment and/or cartilage restoration of the patellofemoral joint as per patient pathology and surgeon discretion. 

Preoperative Planning 

Physical examination is performed to document the extent of pathologic patellar lateral translation,and to test the competence of the medial patellofemoral ligament. Significantly increased abnormal lateral patellar glide (translation) with the knee at approximately 30 degrees of flexion should be present to warrant MPFL reconstruction. This procedure is primarily indicated when there is 2 to 3+ increased patholaxity of the medial restraints, in revision cases where more simple reefing procedures have failed, as well as cases where trochlear dysplasia has led to progressive laxity of the medial restraints. Once again, the surgeon must document MPFL patholaxity by standard physical examination and confirm the same with examination under anesthesia. Arthroscopy will allow staging and treatment in some cases, when there are associated articular cartilage lesions—while some more extensive lesions 

may suggest cartilage restoration may be a necessary concomitant procedure. It is important to use the Fulkerson test to rule out iatrogenic medial dislocations (post excessive lateral release) which can present with the patient reporting lateral dislocations by mistake. Standard anterior/posterior, lateral and merchant radiographic views, in conjunction with a thorough examination, are usually sufficient to plan surgery. Occasionally, stress radiography can be helpful. Note that true lateral radiographic views are useful for classification of trochlear anatomy/ dysplasia and to document patellar infera or alta. MRI and/or CT tracking studies contribute to planning information, especially in assessing tibial tubercle to trochlear groove distance (“TT-TG measurement”), tilt and MRI evidence of chondrosis. Nevertheless, it is worth reemphasizing that PF problems are often multifactorial and the debate as to whether the MPFL reconstruction can address all the problems of patellar instability or if it should be combined with tubercle surgery, or tubercle surgery should be performed alone is an ongoing one that only long-term prospective studies can decide. 

Graft Selection 

The preferred graft for this procedure is a doubled semitendinosus autograft or allograft. As the native MPFL tears at 200 Newtons, this graft offers a wide “margin of safety” in terms of strength and the doubling of the graft is to match the attachment site area at the patella and certainly not for strength. If autograft is used, it is harvested through a separate small incision over the pes group tibial insertion, after EUA and arthroscopy have confirmed the need for MPFL reconstruction. Alternatively, this procedure can also be performed with a doubled gracilis tendon, particularly for smaller patients. Another option, when a hamstring allograft is not available (but allograft is preferred over autograft), is a singled tibialis tendon allograft. The stiffness of a reconstructed MPFL is approximately 12N/mm, which is approximately 20% of the average stiffness recorded for looped 30 mm long semitendinosus grafts (4, 10) . With any of the graft choices discussed above (whether or not a single or double stranded), the stiffness and strength requirements of the native MPFL are easily met. For the doubled graft technique, the graft is first doubled upon itself. Next a running baseball stitch (using #2 Fiberwire) for a distance of 25mm is performed at each free end and the doubled end creating a “Y” shaped graft (Fig.1). The graft tendon is sized at the doubled end (for a single strand, the larger diameter end). 

Fig.1: Doubled semitendinosus graft sutured in a “Y” configuration ready for insertion at the femoral origin site. 

Surgical Approach 

For isolated reconstruction of the MPFL (without the need for arthrotomy or associated procedures), 2 incisions are made over the respective attachment sites: that is, a 3-4 cm longitudinal incision along the proximal medial border of the patella and a smaller longitudinal incision over the femoral attachment site in the saddle area between the medial epicondyle and the adductor tubercle. (Fig.2). Alternatively, one longer medial patellar incision can be utilized to expose both sites, particularly when an arthrotomy is necessary such as with cartilage restoration of the patellofemoral compartment. 

Fig.2: The incisions centered over the patellar and femoral attachment sites of the MPFL. 

The patellar area incision is made first. Dissection is performed along the proximal . of the medial patella to the interval between layer 2 and 3 (between the MPFL and the capsular layer). This interval is bluntly developed medially towards the medial epicondyle, utilizing a curved Kelly clamp.(Fig. 3) The graft should always be placed extrarticularly (superficial) adjacent to the capsule. A 2 cm femoral incision is made over the tip of the clamp when it overlies the saddle between the epicondyle and the adductor tubercle. 

Fig. 3: Dissection is carried out between layer 2 and 3, and tunneled over to the origin on the femur 

If tubercle surgery (medial/proximal or distal) is planned, then it is typically performed through a separate incision unless a long anterior “universal incision” is planned with an extensive cartilage restoration at the same time of proximal and distal extensor mechanism surgery. As the tubercle surgery alters the relative position of the MPFL as well as peripatellar soft tissue tensions, the tubercle procedures should be performed prior to completing the MPFL reconstruction (noting that the tubercle surgery may possibly warrant some degree of conservative lateral release—only to allow neutralization of excessive patellar tilt documented clinically and radiographically). Be aware that in patients with preexisting adequate soft tissue laxity in the lateral retinaculum, a lateral release could create medial subluxation. In general, a lateral release is often not necessary. If the tubercle malalignment is only mildly to moderately abnormal (the TT-TG distance is in the gray area between “normal and abnormal”), and there is question on the extent or need for tubercle surgery, the MPFL reconstruction can be initiated, but using only pins at the attachment sites and suture to duplicate the planned MPFL. Then the potential need for tubercle surgery can be again reassessed. If distal realignment is thought necessary to centralize the patella, it can then be performed and fixed followed by completion of the MPFL reconstruction with fine-tuning of the MPFL lengths with this new patellar tracking position. 

The femoral attachment of the MPFL is identified using the landmarks of the medial epicondyle, the MCL and the adductor tubercle. The femoral attachment of the MPFL resides in the “saddle” between the adductor tubercle and the medial epicondyle. The fascia is incised and a 2.4 mm guide pin (Bio-Tenodesis fixation set, Arthrex, Naples, Fl.) is placed just proximal to the epicondyle and distal and anterior to the adductor tubercle. A #2 suture is wrapped around this guide pin and then is pulled to the patella with the clamp through the same MPFL/capsular interval tunnel and sutured or clamped into the patellar attachment site of the MPFL (which is along proximal 1/2 of the medial patella noting the arms of the suture are at the proximal and distal extent of the patellar MPFL attachment footprint). The suture is pulled so that it is not loose or tensioned with the knee at approximately 30 degrees of flexion. The knee is then placed through a range of motion. The suture should become lax with increasing flexion and minimally change or slightly tighten in terminal extension. The planned attachments sites and MPFL length should not over-constrain, tension or tilt the patella medially at any point during full range of motion. 

The femoral origin is much more sensitive than the patellar attachment in terms of achieving an anatomometric graft placement (that is, the femoral attachment site alters the attachment site distances through range of motion more than the patellar site, as a result of the “cam” shape of the medial femoral condyle). This femoral sensitivity of graft attachment site is somewhat analogous to the femoral attachment site importance in an ACL reconstruction relative to the tibial attachment site.. Slight variations in position of the femoral attachment site can have major implications on the patellar tracking and contact forces (9). If the suture does not exhibit the desired length changes during range of motion, the pin in the epicondyle is repositioned. Usually, if there is excessive tightening in extension, the graft is too distal or posterior, and if there is increased tension in flexion, the femoral point is too proximal or anterior. Distal/proximal movement has the greatest effect. The most common error is over constraining and thus causing abnormal contact forces on the medial facet of the patella, as well as making it more difficult to regain flexion in the post-operative rehabilitation. After an anatomometric position is established, tenodesis at the femoral attachment site is the next step. 

Fig.4: Anatomometric testing extension. Fig. 5: Anatomometric testing in flexion. 

Femoral Socket Preparation 

The doubled femoral end of the graft has been sutured and sized. With the femoral guide pin still in place, a cannulated drill bit from the bio-tenodesis system is chosen that is 0.5mm larger than the graft size. Table 1 provides a guide for details of socket preparation and screw selection. 

Table 1: Guide for bio-tenodesis graft fixation. 

Recommended Graft, Socket and Screw size for MPFL Reconstruction Graft Diameter Socket Diameter Socket Depth Bio-Tenodesis Screw Size55.517 mm5.5 x 15 mm5.5617 mm5.5 x 15 mm66.517 mm6.25 x 15 mm6.5725 mm7 x 23 mm77.525 mm7 x 23 mm7.5825 mm7 x 23 mm88.525 mm8 x 23 mm

The femoral socket is drilled to a depth 2mm longer than the designated screw size length. (Fig.6) 

Fig. 6: Femoral socket preparation. 

The appropriately sized Tenodesis Screw (Table 1) is loaded onto the Bio-Tenodesis driver. A #2 Fiberwire suture loop or #2 Fiberwire Suture Snare is passed through the center cannulation of the driver tip. The sutures extending from the graft are placed through the suture loop. The loop is tightened around the tip of the graft (and or knots of the graft suture) and secured as they exit the Tenodesis Driver Handle. The graft is inserted into the base of the femoral pilot hole with the screw on the posterior aspect of the socket and the tendon exiting anteriorly (Fig.7). 

Fig.7: The femoral screw insertion on the posterior side of the socket. 

Adjustments can be made to place the tendon in different aspects of the tunnel (with respect to the screw) to adjust anatomometric characteristics of graft length, if necessary. The Bio-Tenodesis Screw is then advanced adjacent to the graft until the screw is flush to the cortical bone rim. Security of the graft is tested. The sutures exiting the Bio-Tenodesis screw and graft 

sutures may be either tied together to lock the screw and graft together or cut as per surgeon preference. 

Graft Length Selection and Patellar Fixation 

Option A: Bio Suture Tak “Reverse Loop” Technique 

Two doubly loaded Arthrex 3.0 mm (smaller patellae) or 3.7 mm (larger patellae) Bio-Suture Tak (Arthrex, Naples, Fl.) anchors (Fig. 8) are reloaded with the “Fiberwire loop end rerouted through suture anchor loop”. (Fig.9). Alternatively, one 3.7 mm anchor can be utilized for a single strand graft. The suture anchors are placed in a trough in the medial edge of the patella (from the mid-waist, superiorly) anterior to the articular cartilage. 

Fig. 8: Bioabsorbable suture anchor reloaded “loop to loop”. 

Fig. 9a: Medial border of patella trough created with a burr. 

Fig. 9b: Suture anchor embedded in bony trough which has been refitted with looped Fiberwire. 

Route the two free arms of the allograft tendon in the developed soft tissue tunnel interval from the femoral attachment to the patella (Fig.10). 

Fig. 10: Passage of graft to patella after femoral fixation 

The sutures of the double tail ends of the graft are passed into the loops which have been threaded through the suture anchors loops. After pulling the sutures and graft tails through the soft tissue tunnel, the graft tails are pulled through the suture loops. One tail of the “Y” graft is through the mid-waist loop and the other graft tail is through the proximal loop. The suture loops are temporarily cinched tight around the graft and held with hemostats. This initial graft length trial is set at approximately 30 degrees of flexion with both grafts pulled to length with minimal tension and no laxity. The knee should be brought through a full range of motion, with the graft only slightly tightening between 30 degrees and zero, and becoming progressively lax in flexion. The knee flexion position for the first estimation of the graft length is based on a review of biomechanical studies which together suggest that the attachment site distance (femoral to patellar) is longest near 30 degrees of flexion.(Fig. 11) 

Fig.11: With this cinch loop technique, fine adjustments of graft length are possible independent of attachment site selection. 

Optimum graft placement allows approximately a patella balanced in the trochlea or 2+ medial/ 2+ lateral trochlear quadrants of translation (or symmetrical with the other knee, if unaffected). Once satisfied with the graft length, tie the loops and fold the graft back onto itself and then suture the graft to itself. (Fig. 12, 13) 

Fig.12: After tying the anchor suture loops, additional sutures can be added further securing the graft to the soft tissue sleeve at the medial edge of the patella, and the remaining graft can be 

passed back towards the femoral attachment and sutured to itself. 

Fig. 13: The completed reconstruction utilizing suture anchors. 

Patellar Fixation using bio-tenodesis technique 

This method of fixation is ideally suited for single strand grafts. Using the patellar poles as a reference, the 2.4 mm guide pin is placed above the mid-waist of the patella, but not venturing too far proximally so as to place the socket near the thickest portion of the patella in the center of the ligament origin and advanced transversely across the patella until it just exits laterally to ensure placement parallel with the dorsal surface. Isometric (anatomometric) characteristics should be checked again by placing a temporary suture through the graft at the patellar border under appropriate tension. A suture is then placed through the dorsal patellar soft tissues at the edge of the socket and the same tensioning check is performed as described before, securing the suture temporarily through the graft and holding it with a hemostat. Once determined, the graft is then marked at this point. (Fig. 14) 

Fig.14: The guide pin has been placed, and a temporary suture placed at the edge of the patella is used to determine tension. The corresponding location on the graft is marked and then the graft is measured and cut 17 mm distal to this point. 

The graft is then cut 17 mm distal to this point. This portion of the graft up to the mark should then be sized and trimmed as well as tapered slightly to be between 5 and 6 mm, taking into account that the baseball suture to be placed will slightly bulk up the graft. A graft larger than this would require too large a socket and screw for patellar fixation. A baseball suture is placed in this 17 mm free end of the graft and sized once more. An acorn reamer 0.5 mm larger than the graft size is chosen as per the same guidelines in Table 1 and a 17 mm socket is created. (Fig. 15) 

Fig.15: Drilling the patellar socket 

The appropriate sized screw, most commonly the 5.5 x 15 mm or the 6.25 x 15 mm, is placed on the appropriate driver (Table 1). The graft is pulled up to the tip of the driver with a suture loop through the cannulation of the driver as before 

The knee is placed in 30 degrees of flexion as the bio-tenodesis screw is inserted on the inferior portion of the graft until flush with the cortical rim of the patella. The graft sits superiorly and the screw is in the thicker portion of the patella. Additional backup fixation is possible with a 3.0 mm Bio-suture–Tak placed next to the socket, but this is usually not necessary. Soft tissue sutures through the graft and the patella can also be utilized. (Fig. 16) 

Fig.16: Securing the graft to the patella, placing the screw inferiorly, in the thicker portion of the patella. 

The soft tissues of layer 2 can then be reapproximated over the graft and imbricated if necessary. The remaining wound is closed in standard fashion. 

Post Operative Management 

1-6 weeks post-op 

A compressive soft dressing and range of motion brace, initially locked in extension, are applied. The brace is removed toallow full range of motion exercises early in the post-operative period. The brace is opened to allow full extension to progressive flexion as muscle control is reestablished. Quad isometrics are allowed, unless the surgeon has elected to perform VMO advancement. Early motion is important to prevent excessive scarring. Additionally, a CPM may be used in the post operative period. During ambulation, the use of crutches and a knee brace in extension is recommended until the patient has good quad control without lag. Weight bearing as tolerated is allowed. Crutches are gradually weaned and are discontinued when there is excellent operative extremity control and no limp. Standard patellofemoral proximal and core functional muscular exercises are performed throughout the postoperative period. Close monitoring of motion is essential to assure full range of motion is achieved in the early post operative setting---as this procedure can initiate a vigorous scar response. If 

articular cartilage procedures are performed concomitantly, the progression of exercises, ROM and weight bearing is modified accordingly. 

After 6 weeks post-op 

Progressive strengthening and functional exercises are continued post operatively until full strength, endurance and agility have been reestablished. Once all goals are met, activities are advanced through a functional progression program. Patellofemoral bracing may be used as per patient and surgeon preference. 


Multiple studies agree that the MPFL is the primary static restraint to lateral displacement of the patella (4, 11, 3). An MPFL tear is the "essential lesion" after acute traumatic patellar dislocation (3, 2). In patients with recurrent lateral patellar instability, by definition the MPFL must be attenuated, torn or insufficient. Multiple proximal procedures have been described to treat the MPFL pathology. These procedures include arthroscopic and open approaches, reefing and non-anatomic tethering grafts. Those procedures, which are not anatomic, may create abnormal tracking patterns and abnormal contact forces on the patella. Therefore, an attempt has been made to restore a portion of the normal anatomy without over-constraint. Reconstruction of the MPFL is indicated in patients with recurrent lateral patellar instability due to excessive laxity of the medial soft tissue patellar stabilizers, revision cases, and patients with medial laxity with underlying trochlear dysplasia. Examination under anesthesia and arthroscopy routinely precede the reconstruction, not only to document the laxity, but also to document tracking patterns as well as to stage and treat associated articular cartilage lesions. 

Multiple procedures have been described to reconstruct medial patellofemoral ligament (3, 5, 8, 12). Most of these procedures are performed with a free tendon hamstring graft, which is orders of magnitude stronger than the native MPFL. With any approach, careful review of the relevant biomechanics of the MPFL is a critical aspect of reconstruction so as not to over-constrain the patellofemoral joint, particularly the medial facet. In light of the cam shape of the femur, the MPFL is most sensitive to the femoral attachment point. Therefore, it is critical to pick the appropriate points of attachment and test the anatomometric behavior to plan the final fixation points for the procedure. Elias et al. have show that as little as a 4 mm proximal shift in the femoral attachment point of the reconstructed MPFL significantly increases the compressive forces on the medial facet of the patella, as well as increasing medial patellar tilt (9). Likewise, in the same study, they showed that increasing the tension excessively by a length change of as little as 4 mm in the graft also overloads the medial patellofemoral cartilage. 

A common problem is placement of the femoral origin too proximally near the adductor tubercle, such that the graft tightens in flexion, leading to markedly increased compressive forces on the medial patellofemoral cartilage in flexion, and difficulties in regaining range of motion. Since the MPFL primarily functions from 0-30 degrees (5), it would follow that maximum resistance to abnormal lateral tension should be in this range. Likewise, it is critical that the ligament reconstruction should act not as a tether, but only as a “check rein” to prevent excessive lateral displacement of the patella. Normal physiological glide should be reestablished at the end of the procedure. The currently described technique allows a reproducible method, not only to examine the anatomometric measurements of the MPFL, but also to "fine tune" the graft lengths to reestablish normal physiological patellar tracking. 

Pull out strength studies of the fixation utilized demonstrate that the fixation of the grafts is more than adequate to allow for immediate range of motion after these procedures are performed(1). With secure fixation at both attachment sites and a strong graft tissue, an aggressive early rehabilitation program is not only allowable and safe, but essential to decrease scarring and muscular dehabilitation. 


At this point in patellofemoral understanding, it is important to remain current in the ongoing debate regarding patellofemoral malalignment (malalignment may range from an increased TT-TG distance to abnormal femoral or tibial torsion and the potential effect on the MPFL or MPFL reconstruction (6) . Although the precise anatomometric characteristics of the MPFL graft have not been universally agreed upon in the literature, within the range of reported MPFL characteristics, the currently described technique is reproducible in restoring “near normal” MPFL attachment site positions. This allows the reestablishment of patellar stability without creating abnormal tracking, which is paramount as the MPFL serves as a “check-rein” against laterally directed displacement forces and should not over-constrain the patellofemoral compartment. The adherence to attachment site anatomy and graft length and tensioning allows both correction of the pathophysiology of lateral patellar instability and for an aggressive rehabilitation program. 


1. Amis, A.A.; Firer, P.; Mountney, J., et al. (2003). “Anatomy and Biomechanics of the Medial Patellofemoral Ligament.” Knee 10: 215-220. 

2. Avikainen V.J.; Nikkurk (1993). “Adductor Magnus Tenodesis for Patellar Dislocation: Technique and Preliminary Results.” Clinical Orthopedics 297:12-16. 

3. Burks, R.T.; Desio, S.M., et al. (1998). “Biomechanical Evaluation of Lateral Patellar Dislocations.” American Journal of Knee Surgery 11(1): 24-31. 

4. Conlan, T.; Garth, W.P., Jr., et al. (1993). “Evaluation of the Medial Soft Tissue Restraints of the Extensor Mechanism of the Knee.” JBJS American 75(5): 682-693. 

5. Davis, D.K.; Fithian, D.C. (2002). “Techniques of Medial Retinacular Repair and Reconstruction.” Clinical Orthopedics 402: 38-52. 

6. Dejour, H.; Walch, G., et al. (1994). “Factors of Patellar Instability: An Anatomic Radiographic Study.” Knee Surg. Sports Traumatology Arthroscopy 2(1): 19-26. 

7. Desio, S. M.; Burks, R.T., et al. (1998). “Soft Tissue Restraints to Lateral Patellar Translation in the Human Knee.” American Journal of Sports Medicine, 26(1): 59-65. 

8. Drez, D., Jr.; Edwards, T.B.; Williams, C.S. (2001). “Results of Medial Patellofemoral Ligament Reconstruction and Treatment of Patellar Dislocation.” Arthroscopy 17: 298-306. 

9. Elias, J.J.; Cosgarea, A.J. (2005). “Tension in a Reconstructed MPFL Could Overload Medial Patellofemoral Cartilage.” Manuscript Submission for the Patellofemoral Research Excellence Award & Winner of the Patellofemoral Research Excellence Award 2005. 

10. Hamner, D.L.; Brown, C.H., Jr.; Steiner, M.E., et al. (1999). “Hamstring Tendon Grafts for Reconstruction of the Anterior Cruciate Ligament: Biomechanical Evaluation of the use of Multiple Strands and Tensioning Techniques.” JBJS American 81: 549-557. 

11. Hautamaa, P.V.; Fithian, D.C., et al. (1998). “Medial Soft Tissue Restraints in Lateral Patellar Instability and Repair.” Clinical Orthopedics, 349: 174-182. 

12. Muneta, T.; Sekiya, I.; et al. (1999). “A Technique for Reconstruction of the Medial Patellofemoral Ligament.” Clinical Orthopedics 359: 151-155. 

13. Nomura, E.; Fujikawa, T., et al. (1992). “Anatomical Study of the Medial Patellofemoral Ligament.” Orthopedic Surgical Supplement 22: 2-5. 

14. Ostermeier, S.; Stukenborg-Kolsman, C., et al. (2005). “Function of the Medial Patellofemoral Ligament According to Lateral Patellar Instability / An In Vitro Comparison of Soft Tissue Reconstruction to Medial Transfer of the Tibial Tuberosity.” Manuscript Submission for the Patellofemoral Research Excellence Award, presented 2005. 

15. Sandmeier, H.; Burks, R.T., et al. (2000). “Effective Reconstruction of the Medial Patellofemoral Ligament on Patellar Tracking.” American Journal of Sports Medicine 28: 345-349. 

16. Schock, E.; Burks, R.T. (2001). “Medial Patellofemoral Reconstruction Using A Hamstring Graft: Operative Techniques” Sports Medicine 9: 169-175. 

17. Steensen, R.N.; Dopirak, R.M., et al. (2004), “The anatomy and Isometry of the Medial Patellofemoral Ligament: Implications for Reconstruction.” American Journal of Sports Medicine 32: 1509-1513. 

Dr. Anthony Schepsis Sitemap
Website Builder