The safety and efficacy of monoclonal antibodies for canine OA

In this narrative review of bedinvetmab, all publications to date (August 2025) have been considered in chronological order to enable the reader to see the evolution of understanding associated with the product. We also provided information on the human product, tanezumab which never made it to the market.

Studies supporting the marketing authorisation of bedinvetmab (Librela/Beransa)

In 2021 Corral et al published their findings which supported the European marketing authorisation for Librela (bedinvetmab). In 2023 in support of the North American licensing of the same product further data was requested by the US Food & Drug Administration (FDA). The findings of these two studies are comparable and are detailed in the table below.

Table 1. Comparison of two studies evaluating the safety and efficacy of monthly bedinvetmab in dogs

	Corral (2021)	Michels (2023)	Commentary
Subjects	287 client owned dogs	272 client owned dogs with OA	Sample size appropriate for primary outcome measure
Study design	Double blind, randomised, multicentre placebo controlled study	Double blind, randomised, multicentre, placebo controlled, parallel group study	Similar/near identical study design
Placebo	Saline SC n = 146	Saline SC n = 137
Bedinvetmab	0.5 – 1.0 mg/kg SC n = 141	0.5 – 1.0 mg/kg SC n = 135	Range of doses used according to vial
Duration	Dosing days 0, 28, 56. Last assessment day 84	Dosing days 0, 28, 56. Last assessment day 84	Unclear from manuscript if dosing occurred on day 84
Primary outcome measure	CBPI at day 28 Success defined as >1 reduction in pain severity score and >2 reduction in pain interference score	CBPI at day 28 Success defined as >1 reduction in pain severity score and >2 reduction in pain interference score	Identical primary outcome measures
Primary outcome results	43.5% dogs achieved treatment success at day 28 vs 16.9% placebo	47.4% dogs achieved treatment success at day 28 vs 36.6% placebo	Marked difference in placebo results (20%) – unclear reason for this
Number needed to treat (NNT)	Not calculated	NNT decreased with each dose; 19.6 day 7 to 4.3 day 84.	A 2 week study with carprofen for OA documented an NNT of 5 (Vasseur et al 1995)
Secondary outcome results	50.8% dogs achieved treatment success at day 56 & 48.3% at day 84 vs <25% placebo	55.9% dogs achieved treatment success at day 56 vs 39.8% placebo, and 57.4% at day 84 vs <34.2% placebo	Driver of placebo values uncertain
Additional results	QoL improvement higher in bedinvetmab group at every visit Veterinary clinical assessment was significantly improved in bedinvetmab group (69.2-91.4%) at each visit vs placebo (<57.9%)	Bedinvetmab half life was 19.1 days after first dose QoL improved in a higher proportion of bedinvetmab treated dogs	Supports monthly dosing
Additional detail	89 dogs continued to receive 6 additional doses in an open-label phase – sustained efficacy was demonstrated
Adverse effects (AEs)	MSK disorders most common; joint pain 15.1% placebo vs 2.1% bedinvetmab group. Mild injection site reaction in one dog/group. Other AEs reported to be associated with incidental co-morbidities	Similar between groups and not considered treatment related. AEs reported to be not unusual for a population of older dogs diagnosed with OA.
Rescue analgesia required	Placebo n = 19, bedinvetmab n = 3	Placebo n = 8, bedinvetmab n = 7	Unclear why differences in rescue in placebo group
Number removed from study	Placebo n = 22, bedinvetmab n = 9	Placebo n = 19, bedinvetmab n = 15
Clinical pathology	Mild non clinically significant changes	Values fell within reference ranges at all time points
Immunogenicity	4 dogs developed anti-drug antibodies (ADAs)	1 dog per group developed ADAs. Placebo case determined to be false positive. No effect on efficacy in bedinvetmab case.
Author conclusions	Bedinvetmab is safe and efficacious for the alleviation of pain associated with OA in dogs as assessed by owners, veterinarians and laboratory clinical pathology.	The results further corroborate the safety and efficacy of bedinvetmab administered monthly to dogs for the control of pain associated with OA

*NNT = the average number of patients who need to be treated for one to benefit compared to a control group

Commentary

The primary outcome measure assessed at day 28 was the change in the Canine Brief Pain Inventory (CBPI) scores. Percentages of treatment success also increased by days 56 and 84. The CBPI was also used as the primary outcome measure in the licensing studies for two other OA treatments – grapiprant (Galliprant) and enflicoxib (Daxocox). The CBPI is a validated Client Reported Outcome Measure (CROM) and in each of these studies, was used in accordance with treatment success definitions (Brown et al 2013).

A placebo effect was evident in these studies. How does that placebo effect compare to other recently licensed OA treatments? In the licensing work for grapiprant the primary outcome measure was the CBPI scores at day 28. A positive effect was documented in 48.1% of dogs versus 31.3% in the placebo group. In the recent enflicoxib study, two doses were studied compared to mavacoxib or placebo. A positive reduction in CBPI scores was documented at 6 weeks in 84% enflicoxib 4mg/kg; 80% enflicoxib 2mg/kg; 83% mavacoxib; compared to 53% in the placebo group. Conzemius & Evans (2012) examined the placebo effect with dogs enrolled in the placebo arm of an FDA-approved study. The relationship between subjective (caregiver responses) and objective (force platform gait analysis) patient outcome measures were assessed, and the authors demonstrated a caregiver placebo effect for owners reporting their dog’s lameness in 39.7%. This information puts the results of the studies discussed herein into context.

Many of the anti-NGF studies in humans used patient reported outcome measures.

The primary end points that have been almost universally utilized in

these studies are the Western Ontario and McMaster Universities Arthritis Index (WOMAC) pain and function subscales, which are combined with physician’s

global assessment (PGA) scales in many studies. The WOMAC is a well-validated measure that is widely used by the OA research community; thus, the use of it in the majority of these studies allows for comparison and facilitates meta- analysis of the data. 

Safety Evaluation of Bedinvetmab in Adult Beagle Dogs: Summary of Three Integrated Laboratory Studies

As part of the licensing work for Librela, three integrated laboratory studies were conducted to assess the safety of bedinvetmab in adult (10-12 month old) Beagle dogs. Across all studies, evaluations included daily health monitoring, veterinary assessments, clinical pathology, systemic drug exposure, and anti-drug antibody testing. Bedinvetmab was tested SC at 1x (1mg/kg), 3x (3mg/kg) and 10x (10mg/kg) dose every 28 days for 7 doses. A placebo group received saline

• Study 1 incorporated additional assessments such as electrocardiography, neurological (pre-study, days 91/92 & days 178/180) and ophthalmic examinations, and radiographic monitoring of the appendicular joints (pre-study and day 170).

• Study 2 focused on evaluating T-lymphocyte-dependent immune function.

• Study 3 investigated the safety of 2 week concurrent administration of carprofen, a nonsteroidal anti-inflammatory drug (NSAID), alongside bedinvetmab evaluating effects on the appendicular skeleton or organs (bedinvetmab 1 mg/kg SC on day 0, followed by saline or carprofen 4.4 mg/kg SC daily for 14 days

• 
  

Studies 1 and 3 also included terminal pathology and histopathological evaluations.

Across all studies, no treatment-related adverse effects were observed in clinical, neurological, ophthalmic, or joint evaluations. There were no changes in immune morphology or function, and no adverse effects from short-term NSAID co-administration. Importantly, no treatment-emergent immunogenicity was detected.

Further to this, Krautmann et al state that ‘the results do not suggest that RPOA occurs in dogs’. It should be borne in mind that this statement is based on a 7-dose study in young dogs without pre-existing OA. 

The findings of Krautmann et al (2021) were challenged by Farrell et al (2024) via a freedom of information (FOI) request, due to a discrepancy between radiographic reports and postmortem examination reports. Additional information not included in the original publication was subsequently disclosed. One dog receiving 1 mg/kg bedinvetmab had “focal proteoglycan depletion and mild focal cartilage necrosis in the left ulna and an erosion in the cartilage and degeneration of the right ulna”.  The FOI request revealed ‘the authors of the FOI summary concluded that “a potential relation [of boney changes] to treatment cannot be ruled out”. Farrell et al requested publication of images of cartilage erosion and challenged the conclusions about joint safety based on 6-months’ follow up of five healthy dogs.

In response to the letter from Farrell et al (2024), a reply was issued (Werts et al 2024), summarised here;

·      Background radiographic and histopathological findings were uncovered in joints, both in control and bedinvetmab-treated animals, via extensive analysis beyond what is routinely evaluated in laboratory-based studies in canines. These were determined to be incidental and not related to treatment.

·      RPOA, or an RPOA-like condition remains as a described syndrome in humans lacking a canine equivalent

·      Current data demonstrates substantial positive benefit: risk ratio, relieving pain associated with osteoarthritis in treated dogs based on over 14 million doses distributed globally

Post-licensing studies by the manufacturer

In 2024 Zoetis published findings (Gildea et al 2024) related to treatment and veterinarian satisfaction with Librela. 1932 responses were collected from 375 veterinarians across France, Germany, Italy, Spain, and the UK. The key findings from this publication were

- Average satisfaction score: 8.0/10

- Discontinuation due to dissatisfaction: <1% of veterinarians

- Compliance with monthly dosing: Over 99%

- Overall treatment compliance: 85%

- Reduction in polypharmacy:

  - Before bedinvetmab: 47% of dogs needed >1 pharmacological treatment

  - After bedinvetmab: 31% (statistically significant, p < 0.05)

- Average number of medications per dog:

  - Before: 1.9

  - After: 1.3 (statistically significant, p < 0.05)

- Patient diversity: Used across a wide range of ages, weights, and OA severity levels

A 2024 publication (Reid et al) entitled ‘Measuring the effect of the anti-nerve growth factor antibodies bedinvetmab and frunevetmab on quality of life in dogs and cats with osteoarthritis using a validated health-related quality of life outcome measure: an observational real-world study’ set out to document quality of life improvements in dogs and cats treated with anti-NGF products. The QoL tool used was Vetmetrica which assesses four domains in dogs plus summary scores;

• Energetic & Enthusiastic (E/E)

• Happy & Content (H/C)

• Active & Comfortable (A/C)

• Calm & Relaxed (C/R)

• Summary Scores: Physical Wellbeing (PWB), Emotional Wellbeing (EWB)

Enrolled subjects were 75 dogs treated with bedinvetmab at day 0, 28 and 56 with QoL evaluations conducted on days 0, 14, 28, 56, 63 and 70.

By day 14 significant improvements were seen in dogs in PWB and EWB and all domains except C/R. The authors comment;

‘Subsequently there was a continued improvement in all domains and summary scores (p ≤ 0.001) except for H/C in the dog’.

‘The overall improvement in all domain scores […] A/C in the dog exceeded the previously reported minimum important difference scores for the QOL measure, indicating a clinically significant change’.

The authors concluded that treatment with bedinvetmab produced a significant improvement in the QOL of dogs.

Efficacy versus NSAIDs: a comparison of bedinvetmab to meloxicam

In an attempt to answer the long-held question over superiority over NSAIDs, Innes et al (2025) examined 101 client-owned dogs with appendicular OA treated with either bedinvetmab (n=52) or meloxicam (n=49). The primary outcome measure was the Canine Orthopedic Index (COI). Both groups showed significant reduction in COI from baseline, with a larger reduction in the bedinvetmab group – however this was not statistically significant. Efficacy improved over time in both groups. Of note were the adverse effects (AEs) seen in the study. A greater number of AEs were seen in the meloxicam group; GI disorders (9), musculoskeletal (3), neurological (3), skin (1), behavioural (1), compared to the bedinvetmab group; musculoskeletal (2), skin (1), systemic (1). The conclusions from this study were that both bedinvetmab and meloxicam are effective at treating the pain of OA, with fewer AEs seen with bedinvetmab. The lead author of this paper is a consultant to Zoetis.

A report of musculoskeletal adverse effects

A letter published by Iff et al (2025) reported a dog with bilateral elbow OA treated with a multimodal analgesic regime. Seven months after substituting robenacoxib for bedinvetmab monthly, the dog was diagnosed with severe swelling of the tarsal and carpal joints and carpal instability. Treatment with firocoxib improved comfort levels. Diagnostic imaging revealed proliferative and erosive changes in the stifles and carpi. Suspicion of an immune mediated process was ruled out via unremarkable synovial fluid. The authors state ‘having ruled out other causes of polyarthropathy, the severe and rapidly progressing radiographic manifestations in combination with clinical deterioration make RPOA after treatment with bedinvetmab a potential aetiology in this animal.’ Further, the authors go on to say, ‘the causal relationship between the polyarthropathy and bedinvetmab in the case we report here remains inconclusive but needs to be considered in the absence of immune-mediated or infectious causes.’

A response to this letter (Budsberg et al 2025) articulated concerns that the conclusions drawn lacked sufficient evidence.

Pharmacovigilance findings

Adverse events were retrieved from the Zoetis Global Pharmacovigilance database from February 2021 to June 2024 by Monteiro et al (2025). The authors were either employees of or consultants to Zoetis. Each AE was coded using the Veterinary Dictionary for Drug Related Affairs (VeDDRA) terminology. In this work the frequency of adverse effects are reported per 10,000 treated animals. The paper focuses on the top 20 VeDDRA terms. The highest AE reported was lack of efficacy at 1.7/10,000 to the lowest AE reported, lameness at 0.36/10,000. To put numbers into context in this period, 18.1 million doses of Librela were sold and 17,162 AEs were reported. Classification of AEs demonstrated 8 rare events and all others were very rare events, according to the Council for International Organizations of Medical Sciences (CIOMS). Frequency calculations assume one treated dog per dose sold.

After lack of efficacy, the most common AEs were polydipsia followed by ataxia. Non specific musculoskeletal effects appears 17 on the top 20 list. Reports of polyuria/pollakiuria, polydipsia, and urinary incontinence (PU/PD/UI) were rare but were disproportionately reported with bedinvetmab as compared to other canine products. Ataxia was reported at low levels in the UK, however reports have increased more recently from US markets following publication of a newspaper article.

Challenges highlighted from this work

·      It is impossible to estimate true incidence based on spontaneous reports

·      Pharmacovigilance data has limitations and a variety of biases are inherent

·      Reporting rates vary in different countries

·      Pharmacovigilance is a joint responsibility – more reports are required with accurate information

In the first update to the UK Summary of Product Characteristics, polyuria/polydipsia was added to the data sheet. The current European Medicines Agency data sheet, accessed from here, lists the following AEs;

Uncommon

(1 to 10 animals / 1,000 animals treated):

Injection site reaction (e.g. injection site swelling, injection site warmth)

Rare

(1 to 10 animals / 10,000 animals treated):

Ataxia, Polyuria, Urinary incontinence Anorexia, Lethargy, Polydipsia.

Very rare

(<1 animal / 10,000 animals treated, including isolated reports):

Hypersensitivity reaction (anaphylaxis, facial swelling, pruritus), Immune-mediated haemolytic anaemia, Immune-mediated thrombocytopenia.

Musculoskeletal adverse effects – case details & reporting patterns

The most recent paper published at the time of writing (August 2025) was conducted by Farrell and colleagues (2025). The authors of this work raise concerns regarding musculoskeletal adverse event rates (MSAERs) in dogs treated with Librela. The first part of the paper is a review of reported AEs in the The European Medicines Agency’s EudraVigilance database (2004–2024). A comparison to other products licensed to treat OA (Rimadyl, Metacam, Previcox, Onsior, Galliprant, Daxocox) was performed. The objective was to assess whether Librela’s MSAER rate exceeds that of comparator drugs by >50%. In the second part, 19 cases are presented where the authors suspect a causal link to Librela treatment.

·       A disproportionality analysis documented that ligament/tendon injury, polyarthritis, fracture, musculoskeletal neoplasia, and septic arthritis were reported ~9 times more frequently in Librela-treated dogs than in dogs treated with the six comparator drugs.

·       Case narratives and diagnoses in the EudraVigilance databased often differed from those originally reported by veterinarians due to modifications made by Zoetis, the MA holder.

• An expert panel review of the 19 cases drew the conclusion of strong suspicion of a causal association between bedinvetmab and accelerated joint destruction

The paper’s authors call for large-scale investigations into bedinvetmab’s risks and pharmacovigilance.

In response to this publication, Brunke & Dewey (2025) stated that ‘the parallel with adverse outcomes seen in human anti-NGF trials—particularly rapidly progressive osteoarthritis—should not be overlooked and echo the concerns of Farrell et al over modification of data by the MA holder. It is currently not possible in the UK to report directly to the competent authority (Veterinary Medicines Directorate) and therefore there is potential for translational errors between the MA holder and the EU database.

These authors highlight a lack of clarity in terminology of AE reporting. In the US, a formal dictionary is not available, although in Europe VeDDRA offers standardised terminology. The VeDDRA list can be accessed here.

Read the tanezumab story here

Summary

·       Bedinvetmab effectively controls the pain of OA and improves QoL in dogs.

·       The vast majority of treated dogs do not experience AEs.

·       The main reported AEs are lack of efficacy, polydipsia and ataxia and are classified as rare or very rare.

·       Clear documentation of the adverse effects of anti-NGF antibodies has been shown in humans, although the underlying aetiology is unclear.

·       Mitigation strategies in humans are not clear, and development of these products has paused.

·       Clinical signs & diagnostic imaging findings supportive of joint pathology have been reported in dogs treated with bedinvetmab.

·       Currently there is no clear association with dose, treatment duration or age.

·       It is not possible to predict which dogs may experience such adverse effects.

·       Reports of ataxia could be a manifestation of peripheral neurological injury – baseline examinations and periodic monitoring are important in dogs treated with Librela. The human experience shows that PN is reversible upon cessation of treatment.

·       Veterinarians are encouraged to record clinical findings at each examination and to report all AEs to national competent authorities or the marketing authorisation holder. Reports should specify which VeDDRA term the veterinarian believes to apply to the clinical findings.

References

Brown DC, Bell M, Rhodes L. Power of treatment success definitions when the Canine Brief Pain Inventory is used to evaluate carprofen treatment for the control of pain and inflammation in dogs with osteoarthritis. Am J Vet Res. 2013 Dec;74(12):1467-73. doi: 10.2460/ajvr.74.12.1467.

Brunke MW, Dewey CW. Commentary: Musculoskeletal adverse events in dogs receiving bedinvetmab (Librela). Front Vet Sci. 2025 Jul 16;12:1628681. doi: 10.3389/fvets.2025.1628681.

Budsberg S, Innes J, Berenbaum F, Knesl O, Monteiro B, Werts A. Response to: Iff I, Hohermuth B, Bass D, Bass M (2025) A case of potential rapidly progressing osteoarthritis (RPOA) in a dog during bedinvetmab treatment. Vet Anesth Analg 52, 263-265. Vet Anaesth Analg. 2025 May-Jun;52(3):372-374. doi: 10.1016/j.vaa.2025.03.002.

Conzemius MG, Evans RB. Caregiver placebo effect for dogs with lameness from osteoarthritis. J Am Vet Med Assoc. 2012 Nov 15;241(10):1314-9. doi: 10.2460/javma.241.10.1314.

Corral MJ, Moyaert H, Fernandes T, Escalada M, Kira S Tena J, Walters RR, Stegemann MR. A prospective, randomized, blinded, placebo-controlled multisite clinical study of bedinvetmab, a canine monoclonal antibody targeting nerve growth factor, in dogs with osteoarthritis. Vet Anaesth Analg. 2021 Nov;48(6):943-955. doi: 10.1016/j.vaa.2021.08.001

Farrell M, Adams R, von Pfeil DJF. Re: Laboratory safety evaluation of bedinvetmab, a canine anti-nerve growth factor monoclonal antibody, in dogs. Vet J. 2024 Jun;305:106104. doi: 10.1016/j.tvjl.2024.106104. Epub 2024 Apr 3. PMID: 38580157

Farrell M, Waibel FWA, Carrera I, Spattini G, Clark L, Adams RJ, Von Pfeil DJF, De Sousa RJR, Villagrà DB, Amengual-Vila M, Paviotti A, Quinn R, Harper J, Clarke SP, Jordan CJ, Hamilton M, Moores AP, Greene MI. Musculoskeletal adverse events in dogs receiving bedinvetmab (Librela). Front Vet Sci. 2025 May 9;12:1581490. doi: 10.3389/fvets.2025.1581490.

Gildea E, North C, Walker K, Adriaens F, Lascelles BDX. Use of Bedinvetmab (Librela^®) for Canine Osteoarthritis in France, Germany, Italy, Spain, and the UK: Quantitative Analysis of Veterinarian Satisfaction and Real-World Treatment Patterns. Animals (Basel). 2024 Jul 31;14(15):2231. doi: 10.3390/ani14152231.

Hochberg, M. C. et al. When is osteonecrosis not osteonecrosis?: Adjudication of reported serious adverse joint events in the tanezumab clinical development program. Arthritis Rheumatol. 68, 382–391 (2016).

Iff I, Hohermuth B, Bass D, Bass M. A case of potential rapidly progressing osteoarthritis in a dog during bedinvetmab treatment. Vet Anaesth Analg. 2025 Mar-Apr;52(2):263-265. doi: 10.1016/j.vaa.2024.11.041.

Innes JF, Lascelles BDX, Bell D, Tulloch R, McVey A, Northcott C, Welbourn M, Higgins K, Horakova V, Maddox TW. A randomised, parallel-group clinical trial comparing bedinvetmab to meloxicam for the management of canine osteoarthritis. Front Vet Sci. 2025 Mar 24;12:1502218. doi: 10.3389/fvets.2025.1502218.

Krautmann M, Walters R, Cole P, Tena J, Bergeron LM, Messamore J, Mwangi D, Rai S, Dominowski P, Saad K, Zhu Y, Guillot M, Chouinard L. Laboratory safety evaluation of bedinvetmab, a canine anti-nerve growth factor monoclonal antibody, in dogs. Vet J. 2021 Oct;276:105733. doi: 10.1016/j.tvjl.2021.105733.

LaBranche TP, Bendele AM, Omura BC, Gropp KE, Hurst SI, Bagi CM, Cummings TR, Grantham LE 2nd, Shelton DL, Zorbas MA. Nerve growth factor inhibition with tanezumab influences weight-bearing and subsequent cartilage damage in the rat medial meniscal tear model. Ann Rheum Dis. 2017 Jan;76(1):295-302. doi: 10.1136/annrheumdis-2015-208913.

Michels GM, Honsberger NA, Walters RR, Kira S Tena J, Cleaver DM. A prospective, randomized, double-blind, placebo-controlled multisite, parallel-group field study in dogs with osteoarthritis conducted in the United States of America evaluating bedinvetmab, a canine anti-nerve growth factor monoclonal antibody. Vet Anaesth Analg. 2023 Sep;50(5):446-458. doi: 10.1016/j.vaa.2023.06.003.

Monteiro BP, Simon A, Knesl O, Mandello K, Nederveld S, Olby NJ, Innes JF, Lascelles BDX. Global pharmacovigilance reporting of the first monoclonal antibody for canine osteoarthritis: a case study with bedinvetmab (Librela™). Front Vet Sci. 2025 Apr 24;12:1558222. doi: 10.3389/fvets.2025.1558222.

Reid J, Gildea E, Davies V, Thompson J, Scott M. Measuring the effect of the anti-nerve growth factor antibodies bedinvetmab and frunevetmab on quality of life in dogs and cats with osteoarthritis using a validated health-related quality of life outcome measure: an observational real-world study. Front Vet Sci. 2024 Aug 14;11:1395360. doi: 10.3389/fvets.2024.1395360.

Vasseur, P. B., Johnson, A. L., Budsberg, S. C., Lincoln, J. D., Toombs, J. P., Whitehair, J. G., & Lentz, E. L. (1995) Randomized, controlled trial of the efficacy of carprofen, a nonsteroidal anti-inflammatory drug, in the treatment of osteoarthritis in dogs. Journal of the American Veterinary Medical Association 206, 807-811

Werts A, Reece D, Simon T, Cole P. Re: Re: Laboratory safety evaluation of bedinvetmab, a canine anti-nerve growth factor monoclonal antibody, in dogs. Vet J. 2024 Aug;306:106175. doi: 10.1016/j.tvjl.2024.106175. Epub 2024 Jun 15. PMID: 38885831.

Wise BL, Seidel MF, Lane NE. The evolution of nerve growth factor inhibition in clinical medicine. Nat Rev Rheumatol. 2021 Jan;17(1):34-46. doi: 10.1038/s41584-020-00528-4. Epub 2020 Nov 20. PMID: 33219344.

Xu L, Nwosu LN, Burston JJ, Millns PJ, Sagar DR, Mapp PI, Meesawatsom P, Li L, Bennett AJ, Walsh DA, Chapman V. The anti-NGF antibody muMab 911 both prevents and reverses pain behaviour and subchondral osteoclast numbers in a rat model of osteoarthritis pain. Osteoarthritis Cartilage. 2016 Sep;24(9):1587-95. doi: 10.1016/j.joca.2016.05.015. Epub 2016 May 18. PMID: 27208420; PMCID: PMC5009895.