In vivo evaluation of temperature-responsive antimicrobial-loaded PNIPAAm hydrogels for prevention of surgical site infection

Abstract

Surgical site infections (SSIs) are a persistent clinical challenge. Local antimicrobial delivery may reduce the risk of SSI by increasing drug concentrations and distribution in vulnerable surgical sites compared to what is achieved using systemic antimicrobial prophylaxis alone. In this work, we describe a comprehensive in vivo evaluation of the safety and efficacy of poly(N-isopropylacrylamide-co-dimethylbutyrolactone acrylamide-co-Jeffamine M-1000 acrylamide) [PNDJ], an injectable temperature-responsive hydrogel carrier for antimicrobial delivery in surgical sites. Biodistribution data indicate that PNDJ is primarily cleared via the liver and kidneys following drug delivery. Antimicrobial-loaded PNDJ was generally well-tolerated locally and systemically when applied in bone, muscle, articulating joints, and intraperitoneal space, although mild renal toxicity consistent with the released antimicrobials was identified at high doses in rats. Dosing of PNDJ at bone-implant interfaces did not affect normal tissue healing and function of orthopaedic implants in a transcortical plug model in rabbits and in canine total hip arthroplasty. Finally, PNDJ was effective at preventing recurrence of implant-associated MSSA and MRSA osteomyelitis in rabbits, showing a trend toward outperforming commercially available antimicrobial-loaded bone cement and systemic antimicrobial administration. These studies indicate that antimicrobial-loaded PNDJ hydrogels are well-tolerated and could reduce incidence of SSI in a variety of surgical procedures.

1. Introduction

Surgical site infections (SSIs) are one of the most common causes of unplanned hospital readmission.^(1),(2) While incidence of SSI is about 1–3% on average, rates and severity vary considerably depending on the procedure.^(3)–(5) Treatment can require additional surgical intervention, prolonged recovery, and uncertain outcomes, especially when the infection involves deep tissues, an organ space, or implants.

Pathogens that cause SSIs vary across different procedures but are usually susceptible to first line antimicrobials.⁽⁶⁾ Systemic antimicrobials are a well-established component of SSI prophylaxis, and administration is clinically timed to establish bactericidal levels in vulnerable tissues. However, the bactericidal effect of antimicrobials is not immediate, and sustained supra-MIC concentrations may be necessary to eliminate all contaminating organisms.⁽⁷⁾ Surgical wounds inherently have dead space, blood and fluid collections, injured tissues, and possibly medical devices that contaminating organisms can colonize before systemic antimicrobials or the host immune system can eliminate them.⁽⁸⁾ Surface-attached colonies rapidly become tolerant to systemic antimicrobials as they form biofilm, leading to an SSI.⁽⁹⁾ SSI prophylaxis could be improved by using local delivery, which enables high antimicrobial concentrations, sufficient to kill both planktonic and surface-attached bacteria, to be maintained in the surgical site without incurring significant systemic toxicity or risk of developing resistance.

We have developed an injectable, temperature-responsive, sustained-release hydrogel for prevention of SSI. The hydrogel is based on a polymer of poly(N-isopropylacrylamide-co-dimethylbutyrolactone acrylamide-co-Jeffamine M-1000 acrylamide) [PNDJ] which transitions from an aqueous solution to a semisolid hydrogel as it warms to body temperature. The polymer consists primarily of N-isopropylacrylamide (NIPAAm) repeat units; which impart the material with a lower critical solution temperature (LCST; the temperature above which the polymer precipitates to form a hydrogel).^(10),(11) Gel dissolution is controlled by time-dependent hydrolysis of side-chain lactones which increases the LCST;⁽¹²⁾ while the rates of hydrolysis and release of entrapped drugs are controlled by polyethylene glycol-based Jeffamine M-1000 side chains via control of the hydrogel water content (Supplementary Figure 1).⁽¹³⁾ PNDJ gels capably fill complex wound spaces,^(12),(14) deliver sustained supra-MIC concentrations of antimicrobials with broad spectrum coverage (e.g., tobramycin, gentamicin, vancomycin) in surgical sites for multiple days,^(14),(15) and dissolve following drug delivery.^(12)–(14)

In addition to providing effective sustained release in vivo, PNDJ hydrogels must also be well-tolerated by local tissues, compatible with adjacent permanent implants, and safe from systemic exposure to the antimicrobials and clearance of the polymer. While NIPAAm-based polymers have been a subject of biomaterials research for over 30 years,⁽¹⁶⁾ there are few studies investigating biodistribution and tolerability of these polymers as sustained release carriers.^(17)–(20) Here we address these criteria in reporting: 1) PNDJ biodistribution; 2) systemic toxicity to the polymer and antimicrobials; 3) pharmacokinetics of locally delivered antimicrobials; 4) biocompatibility with soft and hard tissues; and 5) compatibility with tissue healing at bone-implant interfaces, including ingrowth surfaces on an uncemented total hip arthroplasty. We also report effectiveness compared to clinically available approaches against common SSI pathogens to prevent SSI recurrence after debridement in an osteomyelitis treatment model representing a higher clinical challenge than preventing infection in clean surgical wounds.^{(12),(21),(22)}

2. Materials and Methods

2.1. Polymer Synthesis and Characterization

PNDJ polymers were synthesized by free radical polymerization using reagent grade chemicals as previously described (details provided in Supplementary Methods).^(13),(14) For biodistribution studies, two batches of ¹⁴C-labeled PNDJ were prepared by inclusion of approximately 1 mol% of ¹⁴C-acrylamide monomer (American Radiolabeled Chemicals) with 15 μCi of radioactivity per gram. Low molecular weight (LM_W) radiolabeled polymer was synthesized in equal parts dioxane/THF; high molecular weight (HM_W) radiolabeled batch was synthesized in an 80/20 ratio of dioxane/THF. The repeat unit content of all polymers was characterized by ¹H NMR (400 MHz, Varian) in CD₃OD. Molecular weight was determined by gel permeation chromatography (Waters Styragel HR4 THF column, Wyatt MiniDawn multi-angle light scattering detector, Shimazdu RID-10A refractive index detector). PNDJ polymers used in animal studies were sterilized by ethylene oxide and subsequently handled aseptically. A complete list of polymers used is provided in Supplementary Table 1.

2.2. Preparation of Antimicrobial Loaded Hydrogels

Antimicrobial loaded hydrogels were prepared as previously described.⁽¹⁴⁾ Sterile PNDJ polymer was dissolved in aqueous buffer (20 mM acetic acid-sodium acetate pH 4 or PBS pH 7.4; concentration and buffer detailed below) overnight at 4°C and loaded into syringes. Vancomycin hydrochloride and tobramycin sulfate were co-dissolved in diH₂O (generally 8.8 wt% and 13.2 wt% respectively; concentrations detailed below) and loaded into syringes immediately prior to dosing. PNDJ solution was mixed with the antimicrobials by joining the two syringes with a Luer-lock female-to-female coupler and rapidly pushing the contents between them for 10–12 strokes resulting in even distribution of the antimicrobials. Injection of the resulting solution formed antimicrobial loaded hydrogels as the solution warmed to body temperature. Representative in vitro drug release data for gels used in the studies detailed below were collected as previously described⁽¹⁴⁾ and are provided in Supplementary Figure 2. These data are in agreement with our prior reports that in vitro release of tobramycin and vancomycin from PNDJ is primarily controllable by adjusting the Jeffamine M-1000 acrylamide (JAAm) repeat unit content.^(13),(14)

2.3. Animal Studies

NIH guidelines for the care and use of laboratory animals (NIH Publication #85–23 Rev. 1985) have been observed. All animal procedures were performed by trained medical or veterinary surgeons in accordance with protocols approved by the Institutional Animal Care and Use Committee for each institution (list of institutions, anesthesia, and euthanasia protocols provided in Supplementary Methods).

2.4. Biodistribution

Biodistribution of HM_W and LM_W ¹⁴C-PNDJ without antimicrobials was evaluated following intramuscular injection in Sprague Dawley rats. Rats (n = 6 per polymer, males, ~450 g) received approximately 100–150 μL injections of 25 wt% polymer in PBS in both quadriceps and both hamstrings. Blood was collected from all rats at 1 hr, 6 hr, and 1, 2, 4, 8, 24 and 70 days post-dose (gel dissolution time was 28–42 days). After 70 days, local muscle and major organs (liver, kidney, spleen, lungs, brain, and heart) were harvested, weighed, and homogenized for scintillation counting. Organ homogenates and blood samples were digested (Solvable® digestion solution, PerkinElmer) and prepared in a scintillation cocktail (Hionic-Fluor, PerkinElmer) according to manufacturer instructions. Polymer content was determined as a percentage of the injected dose (%ID) by scintillation counting compared to ¹⁴C-PNDJ standard solutions.

2.5. Toxicity and Pharmacokinetics

PNDJ1.14 gel was prepared at 29.4 wt% in 20 mM acetic acid-sodium acetate buffer (pH 4) with tobramycin (3 wt%) and vancomycin (2 wt%). Thirty Sprague Dawley rats (male, 300–400 g, age 9–13 weeks) received subcutaneous injections of PNDJ in the scruff of the neck while anesthetized. Five escalating dosing levels were tested (0.5, 1.25, 2.5, 3.75, 5 g per 300 g; n = 6 per dose group; details in Supplementary Table 3), representing multiples of the expected human equivalent dose (1x, 2.5x, 4x, 6x, 9x) calculated by allometric scaling guidelines.⁽²³⁾ Body weights were measured immediately after dosing and weekly thereafter. After the two lowest dose groups, groups 3–5 were run sequentially to allow monitoring of potential toxicity prior to proceeding to the next higher dose. Two rats were removed immediately from group 5 due to errors in the gel preparation procedure. Blood was collected for 1) serum chemistry analysis of BUN, creatinine, ALT, and ALP immediately before dosing (pre-dose), on days 3, 7, 14, and 57; 2) tobramycin and vancomycin concentrations at 6, 24, 48, 72, and 168 hr. Serum chemistry analysis was performed by Antech Diagnostics (Phoenix, AZ, USA), and tobramycin and vancomycin concentrations were measured by LC/MS at the Texas A&M Veterinary Medical Diagnostic Laboratory (College Station, TX, USA). On day 57, kidneys were harvested, fixed in 10% formalin, embedded in paraffin wax, sectioned, and stained with hematoxylin and eosin. Sections from group 2 (moderate dose, 1.25 g/300 g, n = 3) and group 5 (maximum dose, 5 g/300 g, n = 3) were evaluated by a blinded board-certified veterinary pathologist. Additional studies evaluating systemic toxicity in rats receiving IP injections of PNDJ without antimicrobials and in rabbits receiving multiple IM injections of PNDJ with tobramycin and vancomycin (3.3 g/kg) are described in Supplementary Methods.

2.6. Local Tolerance in Muscle and Joints

PNDJ1.29 gel was prepared at 29.4 wt% in 20 mM acetic acid-sodium acetate buffer (pH 4) with tobramycin (3 wt%) and vancomycin (2 wt%). Nine New Zealand white rabbits (female, ~3 kg) received gel in a quadriceps surgical resection site and an intra-articular injection in the knee. Sterile saline in a quadriceps surgical resection site of the contralateral hindlimb was a negative control. Soft tissue resection was performed by removing 1 g of quadriceps muscle. The surgical site was closed over a 14-gauge IV catheter then 1 mL of gel or saline was injected to fill the dead space. Intra-articular injections were 1 mL of gel or saline in each knee delivered via an 18-gauge needle. Rabbits ambulated with normal mobility postoperatively. Rabbits survived 28, 42, or 56 days (n = 3 each). After euthanasia, dosing sites were macroscopically examined for presence of gel, and tissue samples were harvested from each dosing site for histopathological evaluation of local tissue response. Tissues were fixed in 10% formalin, embedded in paraffin wax, sectioned, and stained with H&E for muscle samples and both H&E and Safranin-O staining for intra-articular samples. Specimens containing bone were demineralized prior to embedding. Sections were scored by a board-certified veterinary pathologist who was blinded to the treatment groups.

2.7. Compatibility with Orthopaedic Implants

2.7.1. Transcortical Plug

PNDJ1.18 and PNDJ1.93 gel was prepared at 30 wt% in PBS with a high (3.1 wt%) or low concentration (0.3 wt%) of gentamicin sulfate. Five New Zealand white rabbits (female, ~3 kg) received press-fit tapered titanium plugs in the lateral aspect of both distal femora and proximal tibiae which were thoroughly coated in PNDJ gel (details provided in Supplementary Methods). One rabbit died postoperatively from aspiration, and hindlimbs from this rabbit were harvested to assess the bone-implant surface prior to any healing. Rabbits survived for 56 days, and each of the specimens (4 per rabbit) were used to prepare ground sections stained with toluidine blue to determine the extent of bone healing at the implant interface.

2.7.2. Total Hip Arthroplasty

PNDJ1.24 gel was prepared at 29.4 wt% with tobramycin (3 wt%) and vancomycin (2 wt%) in 20 mM acetic acid-sodium acetate buffer (pH 4). Three adult hound dogs (female, 24.9 ± 1.04 kg) underwent an uncemented total hip arthroplasty (Biomedtrix BFX with Lateral Bolt Stem, Biomedtrix, Whippany, NJ) with a lateral fixation bolt by a board-certified veterinary surgeon experienced in this procedure. These implants feature porous titanium surfaces on both the acetabular cup and femoral stem that aid the initial press-fit fixation and enable bone ingrowth for long-term fixation. During the procedure, 2–3 mL of gel was placed in the reamed acetabulum prior to insertion of the acetabular cup, 10 mL was filled into the femoral canal prior to insertion of the femoral stem (with 3–5 mL being lost after implant insertion), and 3 mL was injected in the joint space after deep closure (muscle and fascia) over the implant.

Following surgery, the hounds were allowed unrestricted cage activity with leash walking beginning at 21 days postoperatively. All animals were monitored for clinical signs of implant failure for 84 days. Blood for serum chemistry analysis of BUN, creatinine, ALT, and AST was sampled pre-dose and at 3, 7 and 14 days after dosing, and was analyzed by the Texas A&M Institute for Preclinical Studies Clinical Pathology Laboratory (College Station, TX, USA). Serum samples were also collected pre-operatively and at 0.5, 2, 6, 24, 48, 72, and 168 hr postoperatively for tobramycin and vancomycin analysis by the Texas A&M Veterinary Diagnostic Laboratory (College Station, TX, USA) using an LC-MS method. Implant fixation was analyzed by radiographs (14, 42, and 84 days), clinical observation, and retrieval of the implants and associated tissue for histologic assessment at 84 days. After retrieval, the specimens were fixed in 10% formalin, radiographed, and embedded in polymethyl-methacrylate. Both transverse and longitudinal ground sections (14 per animal) were made of the acetabular cup and femoral stem by Histion (Everett, WA, USA). Sections were stained with Stevenel’s blue and van Gieson stains to visualize fibrovascular, bone, and cartilage tissue ingrowth.

2.8. Efficacy against SSI

The osteomyelitis model we used and have previously reported⁽¹³⁾ was adapted from Evans and Nelson⁽²¹⁾ to include a metal implant as a foreign body to consistently establish infections following bacterial inoculation. A 1 cm defect was made in the left radius of twenty-six New Zealand white rabbits, a 1 cm Kirshner wire (K-wire) was inserted inside the devitalized bone segment, and the site was inoculated with 7.5 × 10⁶ CFUs of either S. aureus MSSA (ATCC 49230) or S. aureus MRSA (ATCC BAA-1556) before closing the wound. Infections were allowed to establish for three weeks before debridement was completed by a trained orthopaedic surgeon blinded to treatment groups. The treatment groups were block randomized and included PNDJ1.51 with tobramycin (30 wt% and 3.14 wt% respectively; volume of gel was individualized and sufficient to fill the wound: 400–1000 μL, mean 540 μL; n = 6 MSSA; n = 6 MRSA), low-dose antimicrobial loaded bone cement (Simplex-P) with tobramycin (1 g tobramycin per batch) dosed in a standardized 6 × 12 mm cylinder (ASTM 451–08; n = 7 MSSA), and systemic tobramycin dosed subcutaneously (10 mg per day) for 28 days (n = 7 MSSA).

2.9. Statistical Analyses

Statistical analyses were performed in Prism 8 (GraphPad) and Minitab 18 (Minitab). All numerical data are presented as mean ± s.d. unless otherwise stated. In biodistribution data, differences between PNDJ formulations were analyzed by two-way ANOVA with Bonferroni post-test. All systemic toxicity data were analyzed by two-factor ANOVA with Dunnett’s post-test. Infection treatment data were compared by Fisher’s Exact test. In all cases, significance is reported for adjusted p < 0.05.

3. Results

3.1. Biodistribution

The biodistribution of ¹⁴C-labeled low molecular weight (LM_W; ~30 kDa) PNDJ models the formulations used in other studies in this work, while the high molecular weight (HM_W; ~80 kDa) formulation exceeds the size limit for renal filtration (Supplementary Table 1). The dissolution time for both was expected to be about 28 – 42 days based on JAAm content. At 70 days after injection, the hydrogel was expected to be fully dissolved and the polymer eliminated from the site. Gross evaluation of muscle at the injection site found no sign of remaining material, and tissue appeared normal for both polymers. HM_W and LM_W polymer was detected in all examined organ homogenates and blood except the brain, with no difference in concentrations between the formulations (p = 0.7136), although high variability was observed in local muscle for both (HM_W: 4.62 ± 4.57 %ID; LM_W: 5.90 ± 9.45 %ID; Figure 1a). Outside of the injection site, polymer was most concentrated in the liver (HM_W: 4.77 ± 2.44 %ID; LM_W: 0.76 ± 0.47 %ID) and kidneys (HM_W: 1.12 ± 0.29 %ID; LM_W: 2.53 ± 0.72 %ID). LM_W polymer content in the spleen, heart, lungs, brain, and terminal blood was below 0.1 %ID. Both polymers were present in blood collected in-life at 0.12 %ID or less.

Figure 1.

(a) Biodistribution of ¹⁴C-labeled PNDJ 70 days following intramuscular injection in rats. Differences between high and low molecular weight PNDJ were negligible. Data are shown as mean percent injected dose (%ID) + s.d. (HM_W: n = 5; LM_W: n = 6). (b) Systemic markers of nephrotoxicity (BUN) and hepatotoxicity (ALT) determined in a dose escalation study following subcutaneous injection of PNDJ with tobramycin and vancomycin in rats (serum chemistry data for creatinine and ALP in Supplementary Figure 4). Evidence of mild nephrotoxicity and hepatotoxicity was observed, particularly in the highest dose group. Data are shown as mean + s.d.; normal range values are provided for each marker;⁽²⁴⁾ significant differences (*) are reported where adjusted p < 0.05 (Groups 1–4: n = 6; Group 5: n = 4). (c) Pharmacokinetics of tobramycin and vancomycin measured for Groups 1–5 in the dose escalation study described in (b). Data are shown as mean ± s.d. (Groups 1–4: n = 6; Group 5: n = 4). (d) Representative histological images of kidneys from moderate and high dose groups in the dose escalation study. The central region of the sample from the moderate dose group (Group 2) is characterized by tubular degeneration, inflammatory infiltrates, and tubular regeneration. Tubules with vacuolated lining epithelium (arrows) surround the previously described area. In the high dose group (Group 5), the predominant feature is extensive tubular epithelial vacuolation (arrows). Small areas also feature more pronounced tubular degeneration (circled). Kidney sections stained with hematoxylin and eosin; scale bar = 100 μm. (e) Systemic markers of nephrotoxicity (BUN) and hepatotoxicity (AST) toxicity in canines (n = 3) receiving PNDJ with tobramycin and vancomycin (0.44 – 0.52 g/kg) during total hip arthroplasty (serum chemistry data for creatinine and ALT in Supplementary Figure 5). Serum concentrations of nephrotoxicity markers BUN and creatinine were not different or declined from pre-dose values. Hepatotoxicity marker AST showed a transient increase at 3 days and resolved to baseline by 7 days, while ALT declined from pre-dose levels. Data are shown as mean + s.d.; significant differences (*) are reported where adjusted p < 0.05 (n = 6). (f) Pharmacokinetics of tobramycin and vancomycin released from PNDJ dosed via intramuscular injection in the maximum feasible dose study described in (e). Data are shown as mean ± s.d. (n = 6).

3.2. Toxicity and Pharmacokinetics

The systemic safety of PNDJ hydrogels was evaluated across four studies (Supplementary Table 2). In an initial systemic toxicity study in rats using PNDJ polymer gel without drugs, serum markers for nephrotoxicity (BUN, creatinine) and hepatotoxicity (ALT, ALP) did not increase in any treatment group outside the range of published normal values, except for ALP in male rats which was above normal throughout the study and declined from pre-dose concentrations (Supplementary Figure 3, Supplementary Table 4).⁽²⁴⁾

In a dose escalation study of PNDJ gel containing 3% tobramycin and 2% vancomycin, BUN and creatinine exceeded normal in the highest dose group (Group 5) at 3 and 7 days, returned to normal by 14 days, and returned to baseline by 56 days (Figure 1b, Supplementary Figure 4). BUN and creatinine remained normal in Groups 1–4, but a transient rise in BUN from baseline occurred in Groups 2 and 3 at 14 days. ALT was above normal at 3 days in all dose groups, but declined back into the normal range and baseline values at 14 days. ALP declined in all groups compared to pre-dose measurements, which were near or above published normal values (Supplementary Figure 4).⁽²⁴⁾ In the three highest dose groups, serum tobramycin concentrations remained above 5 μg/mL and 1 μg/mL at 24 hr and 48 hr, respectively (Figure 1c); sustained systemic exposure above 2 μg/mL is thought to be associated with increased risk of nephrotoxicity.⁽²⁵⁾ Vancomycin serum concentrations were disproportionately lower than tobramycin compared to their respective content in the gel, which is consistent with prior findings in local tissue homogenates in a small animal model.⁽¹⁴⁾ Histopathological evaluation of kidney sections from Groups 2 and 5 found evidence of tubular injury (Figure 1d). In Group 2, tubular lumina were difficult to identify in small (≤ 200 μm) hypercellular areas due to loss of tubules and mixed inflammatory infiltrate. Areas of injury contained regenerating tubules with hyperplastic epithelial lining and were often surrounded by tubules containing vacuolated epithelial cells. In Group 5, histopathological changes were similar, although marked widespread tubular epithelial vacuolation was observed in one animal which masked other degenerative and regenerative changes. The average degenerative changes in dose groups 2 and 5 were characterized as minimal to mild with concurrent minimal tubular regeneration (semi-quantitative scores shown in Supplementary Table 5) and were consistent with changes associated with aminoglycoside toxicity.

Systemic toxicity from PNDJ with tobramycin and vancomycin was also evaluated in canines receiving a total hip arthroplasty (Figure 1e, Supplementary Figure 5). Serum BUN and creatinine were unchanged from pre-dose levels; AST increased transiently from pre-dose at 3 days and returned to baseline at 7 days; ALT declined from pre-dose levels.⁽²⁴⁾ Evaluation of drug levels showed serum tobramycin concentration peaked at 3.64 ± 0.84 μg/mL at 2 hr and declined to 0.68 ± 0.16 μg/mL at 24 hr, whereas vancomycin peaked at 2.88 ± 2.08 μg/mL at 6 hr and declined to 0.74 ± 0.41 μg/mL at 24 hr (Figure 1f). Notably, peak levels of the two drugs were nearly proportionate to their loaded concentrations in contrast to the dose escalation study. Additionally, in a final study of systemic toxicity, rabbits receiving multiple intramuscular injections of PNDJ with tobramycin and vancomycin (3.55 ± 0.13 g/kg) corresponding to approximately 4 times the expected human dose via multiple intramuscular injections also did not exhibit signs of toxicity (Supplementary Figure 6).

3.3. Local Tolerance in Muscle and Joints

PNDJ hydrogel containing tobramycin and vancomycin placed in muscle, articular cartilage, synovium, and bone in rabbits showed minimal histological tissue reaction. On gross examination at 28 days, gels were evident only in the quadriceps surgical dead space site, whereas at 42 and 56 days, no material was observed in any site. Muscle adjacent to the resection sites from both saline control and gel groups exhibited histologically normal wound healing at each time point, with no necrosis or excessive fibrosis (Figure 2). Muscle adjacent to the surgical dead space sites in all animals receiving PNDJ was infiltrated with macrophages containing abundant, finely vacuolated cytoplasm, indicative of phagocytosed foreign material. The PNDJ group also showed focal regions of mineralization at 28 days but not at 56 days. Semi-quantitative histopathological scores (scored 0–4 for none, mild, moderate, marked, or severe reaction respectively) in muscle resection sites receiving gel and saline decreased from marked severity (Gel = 4.00; Saline = 3.67) at 28 days to mild (Gel = 1.67; Saline = 2.00) in both groups at 56 days after dosing (Supplementary Table 6). At 56 days, gel was absent from histological sections, and wound healing was observed in all samples consistent with the time from surgery. In the knee intra-articular injection sites, all synovium and underlying soft tissue, and cartilage were histologically normal by Mankin score at 28–56 days after dosing.^(26),(27) The underlying femoral and tibial bone and marrow (not shown) were histologically normal.

Figure 2.

Representative histological sections of sites dosed with saline control (left column) or PNDJ gel (right column) with tobramycin and vancomycin in a surgical dead space site (SDS) in the quadriceps after 28 days (a, b) and 56 days (c, d), and in an intra-articular injection (IA) in the knee after 28 days (e, f). (a, b) Quadriceps skeletal muscle exhibits regions replaced by dense cellular granulation tissue (asterisks) with similar appearance for saline and PNDJ. A void indicating remaining PNDJ gel is visible in (b) in the bottom right of the image, along with a focal area of mineralization (arrow) which may contain a remnant of gel. (c, d) At 8 wk post injection, there is a decrease in granulation tissue (asterisks) within the muscle. In the site receiving gel, there was no remaining mineralized material. (e, f) Cartilage appears histologically normal at 28 days post injection for both saline and PNDJ groups. Skeletal muscle was stained with hematoxylin and eosin; articular cartilage was stained with Safranin-O and fast green counterstain; scale bars = 200 μm.

3.4. Compatibility with Orthopaedic Implants

In the rabbit transcortical plug model (Figure 3a), PNDJ containing high (3.1 wt%) and low doses (0.3 wt%) of gentamicin were compatible with histologically normal bone healing with bone directly in contact with the grooved implant surfaces at 8 weeks after implantation (Figure 3b, Supplementary Figure 7). In the canine THA model (Figure 4a), radiographs performed at 2, 6, and 12 weeks post-operatively demonstrated stable press-fit implants in all three animals throughout the study (Figure 4b). At 12 weeks, histological examination of bone-implant interfaces documented a continuous distribution of tissue filling the ingrowth surface of the acetabular cup and femoral stem. There were appropriate regions of bone ingrowth, fibrous tissue, and cartilaginous metaplasia for 12 weeks post-implantation. Osteoblasts and bone trabeculae indicative of new bone formation and bone remodeling in response to load bearing were also noted (Figure 4c–e). Bone ingrowth was present within textured areas in direct contact with bone, and trabecular bone ingrowth extended to adjacent areas consistent with reports for stable implants at similar times in the absence of any treatment article (Supplementary Figure 8).^(28),(29) We also investigated the local in vivo distribution of PNDJ applied to a THA femoral stem implanted in a cadaver. Using a radio-opaque agent in the gel, radiographs show the hydrogel completely filled the intramedullary space between the implant and bone and covered the entire implant surface (Supplementary Figure 9).

Figure 3.

PNDJ compatibility with bone ingrowth on titanium transcortical plugs. (a) The implantation site was prepared by drilling a pilot hole (3.5 mm diameter, 8 mm depth) in either the proximal tibia or distal femur of rabbits (1, 2). The pilot hole was filled with PNDJ containing gentamicin (3, 4). Titanium transcortical plugs (5) were then coated with PNDJ containing gentamicin so that the entire implant was covered, including the grooved ingrowth surfaces (6) prior to press-fitting the implant into the hole until flush with the bone (7, explanted distal femur). (b) Macroscopic sections of tibia and femur, demonstrating normal bone healing and apposition onto implant surfaces in areas in contact with cortical bone for samples treated with PNDJ gel with high concentration gentamicin (3.1 wt%), PNDJ gel with low concentration gentamicin (0.3 wt%), as well as a PBS negative control 8 weeks after implantation. Tissue was stained with toluidine blue; images were obtained at normal magnification. High-resolution composite micrographs of each sample are provided in Supplementary Figure 7.

Figure 4.

(a) PNDJ hydrogels loaded with antimicrobials dosed during total hip arthroplasty (THA) in dogs. Following acetabular and femoral reaming, the cavities were filled with PNDJ gel containing tobramycin and vancomycin. Implants were then coated with gel and press-fit in place using standard technique. At wound closure, gel was injected through a catheter placed within the joint capsule to fill the wound space. (b) Representative radiographs in one animal at 2, 6, and 12 weeks after implantation demonstrating stable press fit implants without subsidence or osteolysis adjacent to the implants. Plastic embedded histological sections of the tissue ingrowth surfaces on the (c) acetabular cup and the (d) femoral stem demonstrate dense fibrous tissue (blue) and bone (pink) throughout the ingrowth surfaces. Bone is present where expected at points of cortical contact. (e) The interface between bone (B) and implant (I) demonstrates complete infiltration of collagen producing fibroblasts (single arrows). Focal calcification also occurred within this zone. Cartilaginous metaplasia (C) was noted in one area in the gap between bone and implant. Where the implant was in direct contact with cortical bone, few or no fibroblasts were seen. Formation of new bone with osteoblastic rimming was noted (double arrows). None of the sections presented any evidence of infiltration of acute inflammatory cells (neutrophils) or mononuclear giant cells commonly associated with foreign body reaction. Histological sections were stained with Stevenel’s blue and van Gieson stain; images are shown at 40X magnification. Full slide images of three sections from each animal are shown in Supplementary Figure 8.

3.5. Efficacy against SSI

The efficacy of antimicrobial-loaded PNDJ hydrogel was evaluated in an osteomyelitis infection treatment model containing a metal implant (Figure 5).^{(12),(21),(22)} Four of the thirty animals were excluded for reasons unrelated to treatment (one died during debridement, one died from complications of the initial infection, one developed Pasteurella skin infection remote from the surgical site, one was culture-negative at debridement). MRSA infections were less apparent than those caused by MSSA, with only moderate swelling and purulence at the infection site. At 4 weeks following treatment, cultures from PNDJ treated animals with MSSA infections and MRSA infections were sterile in 6/6 and 6/6 animals respectively, and there was no clinical sign of infection. The active comparator groups were less successful against MSSA infections with 5/7 successfully treated with antimicrobial loaded bone cement (ALBC) and 3/7 successfully treated with systemic tobramycin (active comparators not evaluated against MRSA). However, differences in efficacy between groups were not statistically significant, likely due to small sample size (PNDJ vs. systemic tobramycin, p = 0.07; PNDJ vs. ALBC, p = 0.46).

Figure 5.

Efficacy of locally applied PNDJ with tobramycin against S. aureus infections. (a) Osteomyelitis biofilm-based infection prior to debridement. (b) Implantation of low-dose antimicrobial-loaded bone cement with tobramycin and a metal implant. (c) Injection of PNDJ with tobramycin alongside a metal implant after debridement of the SSI. (d) PNDJ with tobramycin eliminated all MSSA (6/6) and MRSA (6/6) osteomyelitis infections. Antimicrobial loaded bone cement (ALBC) with tobramycin and systemic tobramycin were only evaluated against MSSA infections (n = 7 each group).

4. Discussion

These studies have several limitations. First, vehicle controls were not always evaluated to isolate effects of the hydrogel from those of the antimicrobial(s). In the toxicity study evaluating the gel without antimicrobials (Supplementary Figure 3), PNDJ did not cause elevation in hepatic or renal toxicity markers. In addition, a prior study also found no systemic toxicity from PNDJ gel without antimicrobials (Supplementary Figure 10). Future studies should include these controls. A second limitation is that studies involved various combinations of polymer and antimicrobials, and multiple dosing sites. However, all gels tested had similar composition and properties (viscosity, gelation temperature, gel dissolution time) such that differences are unlikely to have affected safety or efficacy outcomes, and drug loading should be associated with the effects observed in each individual study. Although the drug release profile does vary depending on the dosing site, in prior studies, tobramycin and vancomycin were found to have been released over a period of 24–72 hours in muscle, bone, and intra-articular sites with active drug confirmed by microbiological assay of recovered gels and tissues.⁽¹⁴⁾ Safety was evaluated in a variety of dosing sites because PNDJ is applicable to a broad range of surgical procedures. Third, evaluation of systemic toxicity focused on markers of renal and hepatic function based on biodistribution data. Additional evaluation of toxicity and safety to other organ systems is appropriate in definitive toxicology studies before advancing to clinical use. Fourth, pharmacokinetic data in animals, especially smaller species such as rats and rabbits, do not directly translate to humans. Drugs that are principally excreted unaltered in the urine (such as aminoglycosides) are cleared more rapidly in smaller species, leading to differences in toxicity; interspecies conversions are available to estimate equivalent human doses.⁽³⁰⁾ Fifth, tobramycin and vancomycin pharmacokinetics were not evaluated in the absence of PNDJ. Previously Lin et al. reported the half-life of tobramycin in rats to be 52 min when dosed IV at 60 mg/kg,⁽³¹⁾ while Marre et al reported the half-life of vancomycin in rats was 36 min when dosed IV at 10 mg/kg.⁽³²⁾ In comparison, the serum concentrations of tobramycin and vancomycin in the dose escalation study in rats decreased by half approximately every 24 hr (Figure 1c), indicating that the antimicrobials are delivered by controlled release from PNDJ. Finally, the canine study evaluating compatibility with total hip arthroplasty did not include control animals not receiving PNDJ; control animals were not included because outcomes from the procedure are well established in clinical cases.

The systemic safety of antimicrobial-loaded PNDJ hydrogels was evaluated across four studies (Supplementary Table 2) focusing on renal and hepatotoxicity based on the polymer’s biodistribution. PNDJ is expected to be renally cleared as an intact soluble polymer after dissolution, suggesting a risk of nephrotoxicity.^{(18),(33),(34)} In addition, the primary mode of toxicity for tobramycin, gentamicin, and vancomycin is also nephrotoxicity, which is associated with high and/or prolonged systemic exposure.^(35),(36) In the described studies, minimal to mild nephrotoxicity consistent with systemic exposure to aminoglycosides was found in high dose groups,⁽³⁵⁾ and minor changes in liver enzymes were not considered representative of hepatotoxicity although this should be monitored in future studies (Figure 1, Supplementary Figures 4–6). Using interspecies scaling guidelines (Supplementary Table 7),⁽²³⁾ escalating doses in rats correspond to approximately 1, 2.5, 4, 6, and 9 times the expected dose in a 70 kg human (Supplementary Table 3). Similarly, doses in the canine and rabbit studies equate to approximately 1 and 4 times a human equivalent dose respectively. Although systemic drug concentrations also do not correspond directly to humans, peak concentrations are expected to scale on approximately a mg/kg basis since the blood volume fraction of rats, dogs, and rabbits (64, 85 and 56 mL/kg respectively)⁽³⁷⁾ is similar to that of humans (65–70 mL/kg). By this measure, dose multiples were as high as 50x in the rat study, 1.5x in the dog study, and 12x in the rabbit study relative to a possible human dose of 20 mL. Overall, the toxicity and PK data generated in the dog study should be considered most relevant to humans as it was produced in a large animal model receiving a dose (~13 mL) that is comparable to the expected human dose (20 mL). In addition, a Phase 2a clinical study of a local delivery carrier containing a similar aminoglycoside dose to the PNDJ hydrogels evaluated in this work resulted in safe systemic exposure in humans.⁽³⁸⁾

Biocompatibility studies generally indicated minimal tissue reaction to PNDJ with and without antimicrobials. In the local tolerance study, the time course of local tissue healing was consistent with the in vitro gel dissolution time of about 6 weeks (Figure 2). Future studies should continue to include assessment beyond gel dissolution to assess safety of polymer clearance.

Orthopaedic SSIs are among the most common and serious SSIs as many include implanted medical devices that act as foreign bodies for pathogens to colonize and form biofilm.^(39),(40) Therefore, PNDJ compatibility was evaluated with normal tissue healing at bone-implant interfaces in a non-load-bearing transcortical plug in rabbits⁽⁴¹⁾ (Figure 3) and in a canine model of uncemented total hip arthroplasty (THA) using an implant that closely approximates a human THA (Figure 4). PNDJ did not negatively impact bone healing at implant interfaces in both non-load-bearing conditions (transcortical plug) and load-bearing conditions (THA). To our knowledge, these data represent the first report of a water-based sustained release drug delivery vehicle applied along the entirety of the tissue-implant interface, including porous ingrowth surfaces, in a load bearing prosthetic joint. This study provides evidence that PNDJ gels are compatible with application around permanent implants for prevention of SSI in orthopaedic procedures.

The efficacy of PNDJ with antimicrobials against SSI agrees with our prior findings using the same osteomyelitis model indicating that post-debridement high-concentration multi-day local delivery is effective in preventing recurrence of chronic SSI.^{(12),(22),(42),(43)} All animals receiving PNDJ with tobramycin were infection-free after treatment, although the active comparator groups performed better than we anticipated. Nelson et al. found that twice daily intramuscular administration of 3 mg/kg gentamicin (an aminoglycoside similar to tobramycin) was effective against MSSA in 25% of rabbits.⁽²²⁾ However, we saw a slightly higher infection-free rate with systemic tobramycin in our study (3/7), leading to a lack of statistical significance compared to the PNDJ hydrogel group. Still, if these current data are examined alongside our prior report using the same model, we have thus far achieved 100% successful treatment of osteomyelitis infections with 14/14 treated for MSSA (8/8 in our prior work using gentamicin⁽¹²⁾ and 6/6 here with tobramycin) and 6/6 treated for MRSA. Notably, tobramycin-loaded PNDJ was effective against an MRSA strain that has a high tolerance to tobramycin in biofilm culture.⁽¹⁵⁾ While this infection model does not study SSI prevention directly, it is more rigorous than prophylactic infection models in which treatment is applied immediately after inoculation of planktonic microbes.^(12),(21) For pathogens susceptible to the delivered antimicrobial(s), an implant-associated biofilm infection would reasonably be expected to require higher and/or longer sustained local antimicrobial levels to clear the infection than would be required for SSI prevention.

A unique challenge in SSI prevention is the unknown identity, number, distribution, virulence, and antimicrobial susceptibility of potential contamination in a surgical wound during surgery. Based on these and prior data,^(14),(15) tobramycin and vancomycin have been selected for future study because they are effective against the vast majority of SSI-causing pathogens (including both Gram-positive and Gram-negative microorganisms),⁽⁶⁾ whereas individual antimicrobials are not. Moreover, both have also been used for local delivery in orthopaedic applications.^(45)–(47)

5. Conclusion

This work describes biodistribution, safety, biocompatibility, and efficacy evaluations of antimicrobial loaded PNDJ hydrogels for SSI prophylaxis. Here, we determined that PNDJ gels were well-tolerated both locally and systemically when dosed in peritoneal space, muscle, bone, and joints (Figures 1–4, Supplementary Figures 3–8), but high doses of aminoglycoside-loaded PNDJ gels can lead to sufficient systemic exposure to cause nephrotoxicity (Figure 1, Supplementary Figure 4). Hydrogels were also found to be compatible with normal tissue healing and function of orthopaedic implants when applied at bone-implant interfaces under both loaded and unloaded configurations (Figures 3, 4, Supplementary Figures 7, 8). Finally, local antimicrobial delivery from PNDJ gels was effective in preventing recurrence of MSSA and MRSA osteomyelitis. The results suggest that antimicrobial-loaded PNDJ may be superior to local delivery from low-dose ALBC or systemic tobramycin therapy (Figure 5). Together, the safety, compatibility, and efficacy data presented here support the continued study of antimicrobial-loaded PNDJ hydrogels for prevention of SSI. Future work will include generation of definitive toxicology and safety pharmacology data on the entire formulation and PNDJ alone as a new excipient to support advancement to clinical studies.

Data Availability Statement:

The data that support the findings of this study are available from the corresponding author upon reasonable request.