X‐Ray Visualized Synovial Fluid Carbon Dioxide Sensor for Early Detection of Prosthetic Hip Infections

A radiographically visualized implantable dissolved carbon dioxide (CO2) sensor is developed to non‐invasively detect and monitor hip infections. The sensor is based on a pH‐responsive hydrogel as the sensing material, immersed in a carbonate buffer contained in a capsule with a CO2‐permeable membrane. Dissolved CO2 decreases the buffer pH causing the polyacrylic acid‐based hydrogel to shrink. The hydrogel length is determined using plain radiography by measuring the movement of a radiodense tantalum bead embedded in the hydrogel with respect to a metal wire in the casing. The sensor shows a clear response in the range of 15–115 mm Hg CO2, with a 4.3 mm Hg precision at 15 mm Hg CO2 level and a 6.4‐hour response time. The hydrophobic CO2‐permeable membrane is impermeable to aqueous molecules rendering the CO2 measurement independent of the external solution pH, which can be measured with a separate X‐ray visualized sensor. Separating the sensor from the aqueous environment is also expected to reduce the potential for biofouling and increase longevity. In summary, the first dissolved gas sensor that can be visualized radiographically is described; it has potential applications in detecting and studying implant infection and local carbon tissue CO2 levels.


Introduction
Implant-associated infections are a devastating complication of joint replacement surgery with an incidence of ≈0.5-2% of total hip arthroplasties. [1,2]The consequences included prolonged DOI: 10.1002/adsr.202300001hospitalization, multiple operations, significant permanent deformity, or loss of the implant. [1,3,4]Costs can be staggering: direct costs of prosthetic joint infections treatment are ≈$100 000 per episode, [5] and lifetime treatment cost for a 65-year-old is an estimated $390 806. [6]t may result in a higher mortality ≈5% increase within 2 months of diagnosis. [7]][15] The prosthetic hip joint provides a surface for the attachment of microbial cells. [14,16][22] Generally, due to low oxygen levels, anaerobic fermentation in the biofilm leads to local depletion of nutrients and accumulation of metabolic waste products such as lactic acid, citric acid, CO 2 , propionic acid, glycerol, ethanol, etc., within the biofilm (Figure 1a). [17,23,24][27][28] The studies showed a decrease in P O2 was accompanied by a decrease in pH and an increase in P CO2 and lactic acid concentrations. [26,29,30]In a study by Treuhaft and McCarty, a sharp rise in P CO2 and lactate levels and a decrease in pH were observed in samples with P O2 levels lower than 27 mm Hg. [26] The rise in lactate may be due to the changeover of local tissues from mainly aerobic to anaerobic metabolism in anoxic conditions resulting in a decrease in pH.An inverse relationship between synovial lactic acid levels and glucose was determined by  S1 (Supporting Information).Data were digitized from prior literature and replotted here as follows: effective brain tissue pH and arterial pH were measured from 10 dogs (each was measured twice); [31] synovial fluid pH data were from synovial fluid effluents of septic joints from 55 patients; [24] ascites pH data were measured from 21 patients with peritoneal dissemination of gastric cancer. [32]nd-Oleson [24] as a result of the metabolism of glucose to pyruvic acid, followed by conversion to lactic acid under anaerobic conditions as shown in Figure 1a.Therefore, due to the poor removal of products, the P CO2 levels would be higher during infection and can be used as a biomarker of infection.The correlation need not be perfect, however, and measuring them independently may be particularly useful where pH and CO 2 are deliberately altered, for example in bicarbonate therapy.
Figure 1b shows the correlation between pH and P CO2 in several different bodily fluids from different studies. [26,31,32]As explained in the theory section and Figure 2, pH is expected to decrease with log(P CO2 ) (i.e., H 3 O + is proportional to P CO2 ).The intercept and slope is determined by the cation concentration (also ionic strength and associated activity coefficients), and this likely explains the differences between the different fluids.All data fit a logarithmic regression fairly well.
Several approaches have been developed to measure P CO2 mainly based on electrochemical and optical methods.[39] CO 2 sensors have also been developed that are based on the Severinghaus principle, [40] where CO 2 diffuses through a gas-permeable membrane into an electrolyte solution resulting in a pH change that can be measured using various methods. [37]The pH measurements were originally carried out using glass electrodes, however, it is not suitable for miniaturization, and other alternative methods including metal oxide electrodes, and solvent polymeric membrane electrodes require reference electrodes which may suffer from drift. [37,41] are developing a CO 2 sensor for early detection of prosthetic joint infections based on a pH-responsive hydrogel as the sensing material.Arthrocentesis (synovial fluid aspiration) is commonly used to detect/confirm infection when suspected from clinical examination and radiology. [42,43]However, the method is highly sensitive to conditions of fluid aspiration/storage, specifically, CO 2 may escape, resulting in pH drift. [44,45]While arthrocentesis is performed if infection is indicated, it is impractical or contraindicated for screening or serial prosthetic joint infection measurements to monitor treatment because it is performed by a radiologist under fluoroscopy or ultrasound guidance which adds cost and scheduling issues; the procedure is painful, can induce tissue damage and allergic reaction to anesthetics or injected X-ray contrast agent used to confirm needle placement; complications are reported in 1-5% of cases; there is a very small but concerning risk of infecting a previously aseptic joint (with associated liability if test was not indicated). [46,47]Thus, noninvasive measurements of synovial fluid CO 2 measurements would be very helpful for early detection when more conservative treatments are often successful, and for monitoring treatment until eradication.Previously, we developed hydrogel-based pH sensors to measure the pH of synovial fluid for early detection of hip infections and the pH of peritoneal fluid for peritoneal dialysis catheter infections using X-ray imaging. [48,49]Although no fouling effect on the sensor calibration curve, response rate or sensor degradation was observed in solutions of tryptic soy broth bacterial cell culture, bovine synovial fluid, bovine serum, highly oxidative hydrogen peroxide, and copper ion medium, or storage in pH 7 buffer, [50] biofouling maybe a potential concern for an indwelling sensor (especially over long periods (a hip prosthesis may last decades).We modified the pH sensor to determine CO 2 levels in synovial fluid with the added advantage of an improved lifetime by encasing the sensor and fluid in a CO 2 -permeable membrane which is impermeable to aqueous molecules. [51]The pH-sensitive hydrogel is placed in a sodium hydroxide solution enclosed by a CO 2 permeable membrane and exposed to different CO 2 concentrations.Higher CO 2 levels would result in a decrease in the pH of the solution, resulting in a change in the size of the hydrogel, which can be measured by X-ray radiography using the distance between the tantalum bead and the tungsten wire. [37,40]Figure 2a shows a schematic representation of the sensor.

Theory of Synovial Fluid CO 2 Sensor
The sensor is covered with a gas-permeable membrane, which allows the diffusion of CO 2 into the buffer solution without the interference of other liquids.The CO 2 that diffuses in will acidify water as represented in the following equilibria: where K CO2 is Henry's law constant and K 1 and K 2 are the first and second dissociation constants of H 2 CO 3 at 25 °C. [52]Charge balance also applies, so that additional alkaline cation (e.g., from the addition of NaOH or NaHCO 3 ) at a fixed CO 2 pressure, increases overall pH, ionic strength, CO 2 capture, and CO 2 sensitivity: The pH was calculated for different CO 2 levels (P CO2 ) by minimizing the charge from Equation ( 5), subject to Equations ( 1)-( 4), and plotted as shown in Figure 2b for distilled water and 1 mm NaOH.The calculated hydrogel length for different pH values and CO 2 percentages were shown in Figure 2c,d, respectively.The plot is linear on a log P CO2 scale in the range plotted implying that (H + ) is proportional to P CO2 to the power of the slope.As seen in the graph and in agreement with Severinghaus et al., [40] in water solution without sodium, the slope of the pH vs log P CO2 curve is 0.5, that is, (H + ) is proportional to the square root of P CO2 ; while with sodium bicarbonate added to the solution, the slope of the pH vs log P CO2 curve is 1, that is, (H + ) is proportional to P CO2 .The addition of 1 mm base to water, raised the pH and doubled the pH analytical sensitivity to CO 2 (slope of the curve).In our experiment, we use sodium hydroxide as the base.We note that we are assuming an equilibrium between the CO 2 partial pressure and dissolved CO 2 and associated carbonate, bicarbonate, and carbonic acid concentration.The reaction time constant is ≈15 s, which is slow compared to diffusion/reaction in our sensor buffer but may matter in cases where CO 2 is rapidly depleted before equilibrium can be established.The rate is dramatically enhanced using carbonic anhydrase enzymes.

Results and Discussion
[55] They can undergo volume changes in response to changes in stimuli such as pH, temperature, light, ion concentration, or electric field. [53,56][59] In addition, hydrogels are used as sensors and actuators, where the stimulus-sensitive hydrogel acts as the sensing element and the transducer convert the swelling of the hydrogel to the optical or electrical domain. [60,61]Conductometric, amperometric, optical, and mechanical methods have been explored to measure hydrogel swelling.Several studies by Herber et al. demonstrate the use of a pH-responsive hydrogel as a sensing material and a pressure sensor as a transducer to design a carbon dioxide pressure sensor for the diagnosis of gastrointestinal ischemia. [37,60,62]The sensor was miniaturized in his last study, [37] however, it does not discuss a method of detection of carbon dioxide levels in vivo in the developed sensor.
We developed a similar CO 2 sensor that is based on the swelling/ deswelling properties of polyacrylic acid (PAA)-based hydrogel, where we will be measuring the length changes of the sensor in response to CO 2 levels, radiographically.[65] PAA polymer coatings have been previously studied for use in preventing corrosion on titanium and other metallic implants. [66]Poly(ethylene glycol) PEG/PAA devices have also been investigated in a rabbit model for use as corneal implants. [67]PAA hydrogels [68,69] are responsive to pH and swell at high pH and de-swell at low pH, around its effective acid dissociation constant (pK a ) of 5.56. [48]At high pH, the pendant carboxylic acid groups get deprotonated and become negatively charged carboxylate ions (─COO -─). [63,70]Due to increased electrostatic repulsions between bound charges on the polymer chains and increased osmotic pressure, the hydrogel swells at high pH. [66,69]In contrast, at low pH the carboxylic acid groups in the PAA chains of the network do not have any charges resulting in less repulsions between the polymer chains and the hydrogel shrinks at low pH.Calibration of the pH hydrogel sensor shows a linear response between pH 4-8, and effective pK a of 5.56 when fitted to a modified Henderson-Hasselbalch equation with a factor (n = 2.50) to account for a spread in hydrogel pK a .Similar results were observed when the experiment was repeated in bovine synovial fluid in the physiologically relevant pH.The sensor response fit well to an exponential with a 30 min time constant, a linear response between pH 4-8, and 0.05 pH units interobserver agreement.Thus, we expect the length changes in the hydrogel resulting from the subsequent pH changes due to CO 2 variations in the synovial fluid can be detected using Xradiography using the developed sensor.In the sensor design, for radiographic measurements, the pHresponsive hydrogel with an embedded tantalum bead is pinned on one end with a tungsten wire in a polycarbonate groove as shown in Figure 2a.Polycarbonate is used in a wide range of biomedical applications such as in blood oxygenators and blood reservoirs in cardiac surgery products, filter cartridges for renal dialysis, and surgical instruments due to its biocompatibility, glass-like clarity, high strength, and impact resistance. [71,72]he groove is then sealed with parafilm as the CO 2 permeable membrane in order to separate the hydrogel sensor from the external solution in the sensor design.Parafilm is a waterproof, semi-transparent, flexible film composed of a mixture of paraffin waxes and polyolefins which is permeable to gases like CO 2 and O 2 . [73]Recently, a study used drug-loaded parafilm as a remotecontrolled thermoresponsive patch for dermal drug delivery. [73]he gas permeability of parafilm for CO 2 is 1200 cc m −2 d at 23 °C and 50% relative humidity.The CO 2 diffuses through the gaspermeable membrane of parafilm into an electrolyte solution (0.1 mm NaOH) resulting in a change in pH (decrease in pH).Subsequently, in response to the pH change, the hydrogel which is in contact with the electrolyte solution decrease in size (Figure 3).Due to the presence of radiopaque markers in the hydrogel, this change in length can be measured using X-ray radiography.

Sensor Calibration
The hydrogel response to varying CO 2 levels is shown in Figure 4a.The variation was from 7 sensors, each sensor was placed in bicarbonate solutions of varying CO 2 levels and the length of the hydrogel was measured.The change in length was calculated for each CO 2 level and normalized to the length of hydrogel without adding CO 2 .As expected, the length decreased when CO 2 levels increased from 15 to 115 mm Hg CO 2 .As calculated, the resolutions are 4.3, 8.0, 19.8, and 43.4 mm Hg at 15, 45, 76, and 115 mm Hg P CO2 , respectively.While the hydrogel length measurement noise level was essentially independent of P CO2 , and came mostly from sensor-to-sensor manufacture variation, the resolution decreased with increasing P CO2 because of the smaller slope of the calibration curve (Figure 2d).

Sensor Response Time
The sensor response with time was measured by first stabilizing the sensor exposed to P CO2 of 15 mm Hg and then transferring it to a bicarbonate solution corresponding to P CO2 of 115 mm Hg, taking photographs of the sensor with time.The gel length (L fit ) is fit to an exponential expression over time (t): where L max and L min are the maximum and minimum length of gel, respectively; t 0 is the time that the CO 2 pressure was changed;  is the time constant.Figure 4b shows the hydrogel length changing with time and the fit to Equation ( 6) ( = 6.39 h).The sensor length changed over time and stabilized within ≈10 h.When the first data point is excluded,  is 3.8 h, the curve fits better after the initial step (t 0 is 2.31 h shown in Figure S2, Supporting Information).The CO 2 pressure was switched relatively quickly compared to the experiment (hours) and the timing is unlikely to significantly affect the time constant.Instead, we expect that the hydrogel and internal buffer were better buffered at the early time points when the pH was higher, as shown in the titration curve (Figure S3, Supporting Information).Regardless, the CO 2 response rate was acceptable for most physiological infection measurements, which occur over hours per day.If a faster response is needed, the sensor could be made smaller, the membrane thinner/more CO 2 permeable, and the membrane exposed area larger.Conversely, for averaging over a larger period such as a week, the membrane could be made thicker/less permeable/smaller area of contact with the internal buffer, and/or the sensor could be made larger.

Sensor Selectivity and Stability
X-ray Images of both sensors in different pH solutions and the length read were shown in Figure 5a,b, respectively.And the comparison of the CO 2 sensor in pH 8 PBS buffer solution, 1 mm NaOH solution (pH 11), and 1 mm CH 3 COOH solution (pH 3.7) was shown in Figure S4 (Supporting Information).The length change in the pH sensor between pH 8 and pH 11 solution is 7.8 mm (9.3%).This is relatively small because the pH sensor is in saturation of length change, and it corresponds to the length changes in Figure 2c (expected length change is less than 10%).The sensor showed a clear response that is independent of external fluid.Although other gases can be present in the body such as ammonia, they are typically at a low concentration (e.g., the normal range for ammonia level is from 17.8-78.5mol L −1 [74] ) compared to normal venous serum CO 2 concentration (22.4-34.2mmol L −1 [75] ).Since the sensor is covered with a gas-permeable membrane, CO 2 can diffuse into the buffer solution without the interference of other liquids.Therefore, the response of the sensor to changes in pH and salt concentration can only be attributed to the CO 2 .
The PAA-based hydrogels are commonly used in drug delivery, biosensors, membrane, and separation devices due to their biocompatibility and extended life span. [59,64]They are expected to have very low degradation since the crosslinked polymer networks are expected to be highly resistant to degradation within the lifetime of the patient. [63,76]Tantalum has excellent anti-corrosive properties as a result of the stable tantalum oxide protective film formed on the surface of the metal. [77,78]Also, tantalum-based materials are widely used in clinical applications as radiopaque markers and medical implants with no adverse health effects. [77,79]The difference between the pH sensor and the developed CO 2 sensor is that here the hydrogel is encapsulated with a CO 2 permeable membrane which makes the hydrogel insulated from the environment to further avoid interaction with large molecules, while allowing CO 2 to penetrate.In addition, the pH hydrogel used in the CO 2 sensor is equilibrium-based and drift has not been observed after long-term incubation in serum or even harsh oxidative environments.It is expected this would dramatically improve sensor longevity in vivo.
Since the sensor is an entirely encapsulated system, the only concern is the possible release of toxic materials.First, the CO 2 permeable membrane blocks direct contact between bodily fluids and the hydrogel preventing biofouling.Second, the PAAbased hydrogel and tantalum used in the CO 2 sensor are both biocompatible in vivo and widely used in biomedical and clinical applications. [59,64,77,79]Even if it leaks, these biocompatible materials won't have adverse health effects.Therefore, the possible release of toxic materials should not be an issue for the developed CO 2 sensor.

X-Ray Imaging of Synovial Fluid CO 2 Sensor
The synovial fluid CO 2 (hydrogel with radiopaque markers in the polycarbonate casing covered with CO 2 permeable membrane) was allowed to incubate at 76 mm Hg (10% CO 2 ) and 115 mm Hg (15% CO 2 ) CO 2 levels.The sensors were then imaged using X-ray radiography (Figure 6a).The plain radiographs clearly showed the positions of the radiopaque markers.
The synovial fluid CO 2 sensor was then attached to the neck of the hip prosthetic implant (Figure 6b,c1), so that the sensor would be in contact with synovial fluid but away from pressurebearing surfaces.The radiograph clearly showed the implant and the sensor position.The change in length of the sensor can be determined by measuring the distance between the tantalum bead and pinning wire, which were clearly visible on the radiograph.

Limitations
The sensor was only tested in vitro and for future in vivo applications it will need to be tested in a live animal.We recently used an X-ray visualized pH sensor to measure peritoneal pH during infection in a 2-week rat peritoneal dialysis model, [49] and found negligible drift through the period and postmortem.This is consistent with previous studies, where we found the hydrogels to be robust even during long-term incubation in oxidative environments. [48,50]For our CO 2 sensor, we expect that encapsulation with the gas-permeable membrane would further insulate the hydrogel sensor from the environment and minimize the effect on the performance of the sensor.Nonetheless, the gaspermeable membrane is in contact with the surrounding fluid and could become fouled slowing the response rate.
We also have not tested the CO 2 sensor for fatigue after a high number of cyclic loading (10 2 -10 6 cycles) which could be important for showing sensor robustness.However, these experiments would be slow (due to the large cycle number and 6.4 h response time per half cycle) and we do not expect cycling will result in fatigue.Fatigue is usually caused by applying cyclic stress to the hydrogel, and cycling the CO 2 concentration should generate minimal internal stress because the hydrogel can move freely with minimal friction or external stress and the material is usually close to equilibrium.Even when the external P CO2 changes rapidly, the CO 2 diffuses slowly through the membrane, and the internal buffer changes slowly compared to the free hydrogel response time (6.4 h vs 30 min [48] ), thus the pH within the hydrogel is relatively uniform avoiding internal stress.
There could be differences between X-ray imaging systems which can affect image clarity and resolution.Additionally, we tested our sensors in vitro without tissue and the tissue scattering could reduce resolution.However, clinical X-rays give excellent images of hip hardware, especially using a standing X-ray in place of a C-arm or portable unit (where prior results were acquired but these lack anti-scatter grids).The X-ray dose for a unilateral hip radiograph is typically ≈0.7 mSv, and the information gained needs to be balanced against the long-term risks from ionizing radiation.However, the current standard of care already includes X-ray imaging during follow-up, and no additional dose is needed to read the X-ray visualized implanted sensor.If additional X-rays are needed, the dose will be small compared to fluoroscopy for image-guided arthrocentesis.The sensor resolution decreases as the CO 2 level increases (Figure 4a).If more sensitivity is needed, either the bead position would need to be more accurately determined (e.g., with higher precision and more reproducible fabrication), or the response would need to increase (e.g., using a longer hydrogel, adding a mechanical gain mechanism, or altering the hydrogel or internal buffer chemistry).Additionally, the sensor needs several hours to equilibrate with the buffer (Figure 4b; Figure S2, Supporting Information).Response rate can be made more rapid by using a groove with a larger surface area in contact with the internal buffer, making a thinner CO 2 permeable membrane, or changing to a more permeable material.
As an alternative to the parafilm membrane, a polydimethylsiloxane (PDMS) film or tubing can be used as a CO 2 permeable membrane.Compared to parafilm, PDMS is widely used as an optically clear, flexible, inert, non-toxic, biocompatible material, and routinely used as a biomedical implant material. [80]inally, we attached the sensor to the hip prosthesis using cyanoacrylate glue to show where the sensor could fit.In practice, it would need to be permanently attached or integrated into the prosthesis and the sensor profile and placement would be important to maintain its function and prevent motion.Additionally, while we have been focusing on prosthetic hip sensors, similar sensors could be placed on other joints, and if the sensor were miniaturized to fit within a biopsy needle, it could also be injected into the tissue to monitor CO 2 in other applications such as cancer research.

Conclusion
We developed the first X-ray visualized sensor for dissolved CO 2 and attached it to a prosthetic hip for non-invasive measurements in synovial fluid.The sensor consisted of a pH-responsive hydrogel, with an embedded radiopaque marker in a carbonate buffer solution communicating with the external environment through a CO 2 permeable membrane.The sensor responds well in the medically interesting CO 2 range (between 15 and 115 mm Hg) within 10 h.The CO 2 permeable membrane makes the sensor selective to CO 2 and it did not respond to the pH of the external liquid, which could be measured independently with a separate X-ray visualized pH sensor.Also, the sensor is biocompatible and is expected to have a long in vivo lifetime because the hydrogel is only exposed to the internal buffer, avoiding biofouling (only the membrane could have biofouling which may theoretically affect the CO 2 diffusion rate and response time but would not alter equilibrium response).Thus, the sensor can be used to measure local CO 2 levels in the synovial fluid to monitor implant infections.
Synthesis of pH Sensing Hydrogel: The hydrogel was prepared by free radical copolymerization of 10 wt.% AAc and 5 wt.% n-OA as the monomers, 1 wt.%PEGDA (Mn 700) as the crosslinker, and 0.1 wt.% 2-oxoglutaric acid as the photoinitiator, with DMF as the solvent.The photo-polymerization reaction was performed under an inert nitrogen atmosphere using UV irradiation (365 nm) from both sides of the reaction cell for 6 h, and the measured temperature near the UV light was 45 °C.The resulting poly (AAc-co-n-OA) hydrogel films were washed with 70% ethanol to remove any residual monomers, DMF, and hydrate the hydrogel.The hydrogel was washed daily for at least 5 days to ensure the removal of unreacted monomers and initiators in the hydrogel film.Hydrogel samples of length ≈10 mm were transferred to pH 7.4 PBS.
Fabrication of Synovial Fluid CO 2 Sensor: The hydrogel was pinned at one end to a polycarbonate groove with a tungsten wire, and a radio-dense tantalum bead (0.5 mm diameter) was embedded in the other end of the hydrogel.The groove was filled with 0.1 mm NaOH solution, and a tungsten wire was glued to the side edge of the groove with a tantalum bead at the end.Next, the groove was covered with parafilm at the top using a commercially available adhesive (Loctite Superglue-Gel Control).
Sensor Calibration: The hydrogel was placed inside a tube open at both ends, one end was covered with parafilm, and 0.1 mm NaOH was added to the tube and the other end was then sealed with parafilm.Then the tube was immersed in a bicarbonate solution with different CO 2 concentrations including 2%, 4%, 6%, 8%, 10%, 12%, and 15.1%, which corresponded to 15, 30, 45, 60, 76, 91, and 115 mm Hg d CO2 , respectively.Dissolved CO 2 standards were prepared by dissolving sodium hydrogen carbonate in deionized water to prepare a 1 m NaHCO 3 solution which was diluted to yield the desired d CO2 . [81]The images of the sensors were taken after 24 h, and the length of the hydrogels (length between the pinning wire and the tantalum bead) was measured photographically using NIH ImageJ software.
Sensor Response with Time: The sensor was the same setup as above.The tube was first placed in a sodium bicarbonate solution corresponding to 15 mm Hg d CO2 (2% CO 2 ) and allowed to equilibrate.Then it was transferred to a bicarbonate solution corresponding to 115 mm Hg d CO2 (15.1% CO 2 ) and images were taken.The hydrogel length was measured using NIH ImageJ software.
Sensor Selectivity and Stability: The sensor (the hydrogel with radiopaque markers encapsulated in a polycarbonate groove and sealed with parafilm) was placed in different external solutions of water, acetic acid, and sodium hydroxide and placed inside a cell incubator.The CO 2 sensor along with the pH sensor was put into different solutions: pH 8 PBS buffer solution, 1 mm CH 3 COOH solution (pH 3.7), and 1 mm NaOH solution (pH 11), then imaged using X-ray radiography (NEXT Equine DR II portable digital radiography system, Carlsbad, CA, with a battery-powered veterinary X-ray generator, Oberhausen-Germany).The lengths of the hydrogels were measured using NIH ImageJ software.
X-Ray imaging of Synovial Fluid CO 2 Sensor on Hip Implant: The synovial fluid CO 2 sensor was placed at the neck of the hip prosthesis and then X-ray images were taken; the hydrogel length (as indicated by tantalum bead position) was measured using NIH ImageJ software.
Statistical Analysis: NIH ImageJ software was used for all the length measurements.The data points in the graph represent mean length measurements for n =7 hydrogel samples and error bars represent the standard deviation of the mean.

Figure 1 .
Figure 1.a) Schematic diagram of cellular glucose metabolic pathways resulting in acidic metabolites.b) Effective brain tissue pH (triangles), arterial plasma pH (dots), synovial fluid pH (crosses), and ascites pH (diamonds) were plotted against arterial CO 2 pressure (P CO2 ).pH is fit to a logarithmic function of P CO2 except for synovial fluid pH (plotted as a linear function as in the original paper; a version with all on a logarithmic curve is shown in FigureS1(Supporting Information).Data were digitized from prior literature and replotted here as follows: effective brain tissue pH and arterial pH were measured from 10 dogs (each was measured twice);[31] synovial fluid pH data were from synovial fluid effluents of septic joints from 55 patients;[24] ascites pH data were measured from 21 patients with peritoneal dissemination of gastric cancer.[32]

Figure 2 .
Figure 2. a) Schematic diagram of the hydrogel-based sensor to measure CO 2 levels.b) Calculated pH versus P CO2 in distilled water and 1 mm NaOH.c) Hydrogel length versus pH based on data and fit from Ref. [48].d) Calculated hydrogel length versus CO 2 percentage and P CO2 in 1 mm NaOH based on curves (b) and (c); inset shows plot on a logarithmic scale.

Figure 4 .
Figure 4. a) Calibration graph of hydrogel-based CO 2 sensor.The length measurements were normalized to the length measurement without adding CO 2 .The data points in the graph represent mean normalized length measurements for n = 7 hydrogel samples and error bars represent the standard deviation of the mean.b) Hydrogel length versus time for the hydrogel-based CO 2 sensor upon raising P CO2 from 15 mm Hg to 115 mm Hg.

Figure 5 .
Figure 5. Dual P CO2 and pH sensing, and insensitivity of P CO2 sensor to external pH.a) X-ray image of synovial fluid CO 2 sensor and adjacent pH sensor at fixed P CO2 and two different pH solutions: pH 8 (PBS buffer solution) and pH 3.7 (1 mm CH3COOH solution).b) Length (mm) between tantalum bead and pinning wire of X-ray visualized CO 2 hydrogel sensor and adjacent pH sensor in pH 8 and pH 3.7 solutions.In supporting information, pH 11 is also shown in Figure S4 (Supporting Information), with hydrogel slightly longer than at pH 8.0 due to saturating size.

Figure 6 .
Figure 6.a) X-ray image of synovial fluid CO 2 sensor at two different CO 2 levels.b) Photograph of synovial fluid CO 2 sensor on the hip prosthesis.c) X-ray image of synovial fluid CO 2 sensor on the hip prosthesis.c1) The zoomed image of the sensor with the radiopaque markers.