Pharmacokinetics of PP353, a formulation of linezolid for intervertebral disc administration, in patients with chronic low back pain and Modic change Type 1: A first‐in‐human, Phase 1b, open‐label, single‐dose study

Abstract

Background

Bacterial infection of the intervertebral disc is difficult to treat because the tissue is usually not vascularized and systemic antibiotic therapy may not reach optimal antibacterial exposure. Here we characterize the safety, tolerability, and pharmacokinetics of PP353, a suspension of micronized linezolid, formulated for direct intervertebral disc administration.

Methods

The safety, tolerability, and pharmacokinetics of an intradiscal administration of PP353, was assessed in Part A of a Phase 1b study and consisted of a single injection of study drug (3 mL of PP353 and 150 mg linezolid). Clinical assessment included initial safety and tolerability of PP353 with continued follow‐up for 12 months. Assessment of linezolid concentration in plasma samples enabled characterization of the pharmacokinetics. Deconvolution of systemic linezolid was used to estimate intervertebral disc linezolid concentration.

Results

Intradiscal administration of 3 mL of PP353 (linezolid 50 mg/mL) to the nucleus pulposus was well tolerated with no reported study treatment‐related severe or serious adverse events and resulted in an average geometric mean linezolid plasma C _max of 1300 ng/mL at 7.27 h post‐administration. The linezolid plasma C _max observed with intradiscal PP353 is approximately 10% that observed with a standard oral or iv administration of 600 mg linezolid. Pharmacokinetic deconvolution estimated that a single dose of PP353 (150 mg linezolid) provided intradiscal bactericidal concentration of linezolid for 96 h and bacteriostatic exposure for up to 120 h after dosing.

Conclusion

Intradiscal administration of 3 mL of PP353 is well‐tolerated and based on the pharmacokinetics following a single injection, a two‐dose regimen of PP353 (150 mg linezolid) on Day 1 and Day 5 ± 1 was selected to explore safety, tolerability, pharmacokinetics, and efficacy in Part B of the Persica 002 study.

A first‐in‐human Phase 1b study of a formulation of linezolid for intradiscal administration to treat chronic low back pain associated with Modic Type 1 changes demonstrates consistent pharmacokinetics enabling selection of a dose regimen to assess efficacy.

1. INTRODUCTION

Bacterial infection of the intervertebral disc, for example, spondylodiscitis, is difficult to treat because the tissue is usually not vascularized and systemic antibiotic therapy may not reach optimal antibacterial exposure within the disc space. ¹ Acute infection can spread to adjacent vertebral bodies and can cause abscesses and, in severe cases, septicaemia. Spondylodiscitis is most frequently caused by Staphylococcus aureus (39%–75%). ² , ³ , ⁴ , ⁵ , ⁶ Chronic bacterial disc infection has also been associated with chronic low back pain (CLBP) and Modic changes Type 1 (MC1). ⁷ The most common bacterial species isolated from disc tissue samples obtained from patients undergoing surgery for CLBP is Cutibacterium acnes (formerly Propionibacterium acnes) found in about 36% of disc tissue samples. ⁷ , ⁸ , ⁹ , ¹⁰ , ¹¹ , ¹² , ¹³ , ¹⁴ , ¹⁵ , ¹⁶ , ¹⁷ , ¹⁸ Oral antibiotic treatment, 100 days of amoxicillin or co‐amoxiclav, in patients with CLBP and MC1 has been evaluated in two randomized controlled trials (RCTs), with 12‐month follow‐up, but prolonged use of high dose oral antibiotics is poorly tolerated with significant gastrointestinal (GI) disturbance. ¹⁹ , ²⁰

An optimal formulation would target the site of the infection directly and require a shorter treatment course. The development and characterization of PP353, a suspension of micronized linezolid in a thermosensitive poloxamer 407 gel that contains iohexol, a radiopaque contrast agent, which enables image‐guided administration into the disc, has been reported. ²¹ PP353 has been developed to provide a treatment for bacterial infection of the intervertebral disc, in patients with CLBP and MC1.

Linezolid is poorly soluble in aqueous solution (2–3 mg/mL), but in PP353 at high concentration (50 mg/mL) forms a homogeneous suspension of mostly insoluble drug substance in a poloxamer 407/iohexol vehicle. ²¹ PP353 represents a formulation at the maximum feasible dose concentration. Previous intradiscal injection studies provide a range of injected volumes of 0.5–4 mL and a dose volume of PP353 of 3 mL was selected, providing 150 mg of linezolid. ²² , ²³ , ²⁴ , ²⁵ , ²⁶ , ²⁷ , ²⁸ , ²⁹ , ³⁰ , ³¹ , ³² , ³³ The provision of the dose of PP353 in 3 × 1 mL syringes provided greater mechanical advantage for administration of the suspension. After each 1 mL dose, administrators considered whether to continue to dose based on patient feedback and review of fluoroscopy images.

A dose‐escalation approach, for example, cohorts with 1, 2, and 3 mL administered, was not adopted as the subjects recruited into the trial have CLBP associated with MC1 changes and therefore potentially have a chronic bacterial infection within the disc space. A lower concentration than 50 mg/mL linezolid or a lower volume than 3 mL would result in a shorter linezolid exposure resulting in a potentially suboptimal dose of the antibiotic. The absence of any significant findings in the preclinical sheep local tolerability study, the low expected systemic exposure of linezolid, and the precedented volume of injection supported the starting dose of 3 mL of 50 mg/mL linezolid in PP353. There was no plan to increase the injection volume or the concentration in this study.

Ethics and regulatory approval have been obtained for the use of PP353 in a Phase 1b clinical trial, Persica 002 (NCT04238676), “The Modic Trial,” in the United Kingdom, multiple European countries, and in New Zealand. The study will investigate safety, tolerability, pharmacokinetics, and efficacy in subjects with CLBP and MC1 or mixed MC1 and Modic 2 (MC2). The study is divided into two parts. The first, Part A, a first‐in‐human open‐label single‐dose study, has investigated how intradiscal administration of PP353 was tolerated and characterized the pharmacokinetics of PP353. This part allowed the selection of a two‐injection regimen to provide at least 8 days disc exposure for part B of the study. Part B, is a double‐blind randomized placebo‐controlled study that will investigate safety, tolerability, and efficacy and is due to complete in late 2024. Here we report the preplanned primary analysis of the systemic pharmacokinetics of intradiscal administration of PP353 from Part A and the selection of the recommended dose of PP353 for Part B.

2. METHODS

2.1. Patients

This part of the Phase 1b study is a single dose, open‐label cohort exploring safety, tolerability, and pharmacokinetics. The study flow chart is shown in Figure 1. The study was registered at www.clinicaltrials.gov (reference NCT04238676). The key eligibility criteria were male and female patients, 18–70 years of age, with CLBP of >6 months' duration in the area associated with a radiological finding of endplate edema of the vertebral body (MC1) or both edema and fat (MC1 and MC2) at a single lumbar level. At screening and pre‐randomization, a numerical rating scale for low back pain (LBP NRS 0–10) using the average reported result from three questions (the current LBP (0–10), the mean LBP during the last 2 weeks (0–10), and the worst LBP during the last 2 weeks (0–10)), was recorded. ¹⁹ , ³⁴ Eligible patients were required to have a LBP NRS of ≥4 if taking chronic pain medication and ≥6 if not, and LBP being greater than their leg pain. Disability was assessed with a modification of the Roland‐Morris Disability Questionnaire, RMDQ‐23 (0–23) which records the impact of the low back or leg pain on the patient, and eligible subjects scored ≥9. ¹⁹ , ³⁵ , ³⁶ Subjects were excluded if the target lumbar disc for injection had lost more than half of its original height based on comparison with adjacent discs, there was a clear alternative cause of back pain, an interventional back procedure had taken place within 6 months prior to screening, or antimicrobial therapy had been previously used to treat the CLBP. Laboratory and clinical data that showed evidence of generally good health were required. Inclusion and exclusion criteria are provided on www.clinicaltrials.gov. Magnetic resonance imaging (MRI) using qualified 1.5 T scanners with T1, T2, and STIR sequences across the T8‐S3 Sagittal plane with a field of view of >40 cm and Dixon Sagittal and Coronal cover across T11‐S3 with field of view >30 cm, in craniocaudal direction (4 mm) slice thickness and 10% gap, was used to assess radiological findings. Modic changes were characterized as MC1 or mixed MC1 and MC2 by a central reader to confirm eligibility for enrollment. ³⁷

FIGURE 1.

Study flow chart.

The study was approved by the UK Regulatory Authority, the Medicines and Healthcare Products regulatory Agency (MHRA), (April 15, 2019), and by the Yorkshire & The Humber—Leeds East Research Ethics Committee (May 24, 2019) and is being conducted in full conformance with the International Committee on Harmonisation (ICH) Good Clinical Practice (GCP), the principles of the Declaration of Helsinki, and the laws and regulations of the country in which the research is conducted, whichever affords the greater protection to the individual. The study complies with the requirements of the ICH E2A guideline (Clinical Safety Data Management: Definitions and Standards for Expedited Reporting). The study is funded by the Sponsor, Persica Pharmaceuticals Ltd. (UK), and clinical trial operations are managed by Micron Research (UK). Potential patients received a patient information booklet on CLBP, Modic changes, and the association with bacterial infection. Three patients were enrolled in Part A of the study from two clinical sites, The NIHR Lancashire Clinical Research Facility at the Lancashire Teaching Hospitals NHS Foundation Trust, Preston, UK, and the NIHR Coventry and Warwickshire Clinical Research Facility at the University Hospitals Coventry and Warwickshire NHS Trust, Coventry, UK. All enrolled patients gave their written consent. Patients were dosed sequentially with a gap of at least 14 days between each subject being injected. The pharmacokinetics of PP353 was assessed after dosing the first three subjects to decide if the data were adequate to select a dose for Part B or whether another three subjects had to be enrolled. Between January 20, 2020 and July 8, 2021, nine subjects were screened and three were enrolled in the study. Data were collected by investigators and entered into an electronic data capture system. The Part A pharmacokinetic data were analyzed by KinetAssist (UK). As only three patients were enrolled in Part A, no characteristics of the patient population are provided, and average data are reported to ensure confidentiality and subject anonymity.

2.2. Study treatment

The subjects received a single intradiscal administration of 3 mL of PP353, a suspension of micronized crystal form II linezolid powder (50 mg/mL), via a double needle technique. Briefly, PP353 is prepared extemporaneously by combining 4.8 mL of cold PP353‐B, a suspension vehicle containing poloxamer 407 and iohexol, with PP353‐A, a vial containing 253 mg of linezolid powder, and shaking briefly by hand, and used within 3 h. ²¹ All subjects were permitted to receive prophylactic antibiotic coverage for the intradiscal procedure as per standard institutional policy with the exception that linezolid could not be used. The procedure was performed under local anesthesia with sedation if required. All subjects were fasted for solids for 6 h and water for 2 h prior to the intervention. Intradiscal administration was performed in the lateral decubitus or prone position. The skin injection site was cleaned and draped as per the local protocol and a suitable local anesthetic was administered in the skin and subcutaneous tissue. Subjects were conscious, alert, and cooperative during placement of the guide needle placement (18‐G Quincke point luer‐lock spinal needle of at least 90 mm), via an off‐midline approach, up to the annulus of the disc to minimize the risk of injury to the nerve root. Multiple‐plane fluoroscopy confirmed the position of the guide needle. PP353 was taken up into 4 × 1 mL luer‐lock syringes. One of the 1 mL PP353 syringes was used to prime a 10‐cm luer‐lock stiff wall extension tube and then it was attached to the administration needle. The administration needle (22‐G Quincke point luer lock spinal needle, up to 178 mm) was advanced through the guide needle and the tip of the needle positioned in the center of the nucleus pulposus, as confirmed by multiplane fluoroscopy. Three, 1 mL doses of PP353 were administered via the extension tube with fluoroscopy after each dose to ensure the correct placement of the needle and to ensure that the dose remained in the disc and did not leak into adjacent tissue. Once the dosing was complete, the needles were removed, and the injection site was cleaned and dressed. Subjects remained in a recumbent position for up to 8 h post‐administration with overnight monitoring at the clinical research facility. The patients were followed up by telephone and clinic visits for 12 months to assess safety and tolerability.

2.3. Bioanalytical methodology

Samples (50 μL) of human plasma (EDTA) containing linezolid and internal standard, deuterated linezolid, were extracted using a protein precipitation procedure and analyzed by an HPLC equipped with an AB Sciex 4000 mass spectrometer. Positive ions were monitored in the multiple reaction–monitoring (MRM) mode. Quantification was by peak area ratio. The lower limit of quantification was 1 ng/mL linezolid.

2.4. Pharmacokinetics

Blood samples were taken at pre‐PP353 dose (−0.5 h) and after dose at 0.5, 2, 4, 8, 12, 24, 30, 96, 144, 192, and 240 h. Plasma was prepared by centrifugation and frozen at −20°C. Plasma linezolid concentration was evaluated using a validated LC–MS/MS method at ACM Bioanalytical Services (UK) (formerly ABS Labs). Noncompartmental pharmacokinetic parameters were estimated using Phoenix software version 8.3 (KinetAssist, UK).

Plasma concentration profiles of linezolid following a single dose of 3 mL PP353 (150 mg linezolid) into the disc of three subjects were used for the deconvolution. ³⁸ Plasma concentrations of linezolid were available up to 144 h for two subjects and to 96 h post‐injection for one subject.

The characteristic profile was obtained from the linezolid monograph ³⁹ : after a 30‐min intravenous infusion of linezolid (600 mg), clearance (CL) was 7 L/h and the volume of distribution (V) was 47 L. The deconvolution was carried out using Phoenix WinNonlin Version 8.3. Linezolid released from the disc was assumed to reach plasma without appreciable sequestration into the surrounding tissues. Preclinical and clinical distribution studies support this assumption. ⁴⁰ , ⁴¹

3. RESULTS

3.1. Pharmacokinetics

Part A of the Persica 002 study is fully recruited (3 patients) and plasma samples were collected over 10 days after one intradiscal injection of PP353 (3 mL, total dose 150 mg linezolid).

Administrations of 3 mL PP353 to the nucleus pulposus were well tolerated with no reported severe or serious adverse events. No significant extravasation was reported from the disc into adjacent tissues. Transient pain was reported by some subjects but this was reported to be in line with the discomfort observed with other non‐provocative intradiscal injection procedures.

Administration of 3 mL of PP353 resulted in an average geometric mean linezolid plasma C _max of 1300 ng/mL at 7.27 h post‐administration (Table 1 and Figure 2).

TABLE 1.

Summary of linezolid human pharmacokinetic parameters for a 3 mL intradiscal dose of PP353 in Persica study 002.

	C max	t max	AUC0‐t	λz	t ½	AUC0‐∞	CL/F	Tdur
	(ng/mL)	(h)	(h × ng/mL)	(1/h)	(h)	(h × ng/mL)	(mL/h)	(h)
Geometric Mean	1300	7.27	25 600	0.0534	13.0	25 700	5840	101
Geometric CV%	18.4	60.1	31.6	6.53	6.53	31.5	31.5	11.0

Abbreviations: AUC(0‐¥), AUC zero to infinity; AUC(0‐t), Area under concentration‐time curve to last measurable time; CL/F, apparent clearance; C _max, maximum observed concentration; CV, coefficient of variation; l_z, estimate of the terminal elimination rate constant; t _1/2, half‐life; Tdur, time from administration for plasma linezolid concentration to fall to 5 ng/mL; T _max, time of peak level C _max.

FIGURE 2.

Mean ± SD linezolid plasma concentrations of PP353 following a single intradiscal injection of 3 mL of PP353 to three subjects. 144 h timepoint is a mean of n = 2.

The linezolid plasma C _max observed with intradiscal PP353 is approximately 10% of the 13 μg/mL observed with a standard administration of 600 mg linezolid. ⁴²

3.2. Deconvolution of PK

The mean cumulative input (±SD) of linezolid into plasma (from the disc), derived from the three subjects, is illustrated in Figure 3.

FIGURE 3.

Mean (±SD) cumulative input of linezolid into plasma after intradiscal administration of 150 mg PP353 to three subjects.

On average, the deconvolution analysis indicates greater than 100% release from the disc; this overestimation is most probably due to interindividual variability associated with using historical iv data to provide the characteristic data.

However, the analysis does suggest that all the administered dose (150 mg) was accounted for after administration to the disc and is consistent with almost all linezolid released from the disc at the time that plasma linezolid falls below 5 ng/mL in plasma.

The amount of linezolid remaining in the disc over time was estimated (Figure 4).

FIGURE 4.

Mean (±SD) amount of linezolid remaining in the disc after intradiscal administration of 150 mg (adjusted to 100% bioavailability).

At 96 h after dosing it is estimated that approximately 440 μg of linezolid remained in the disc. Assuming a disc nucleus pulposus volume of 10 mL, ⁴³ and assuming equal distribution of linezolid throughout the nucleus pulposus, there would be 44 μg/mL linezolid in the nucleus pulposus. The minimum inhibitory concentration (MIC) of linezolid is 1 μg/mL and thus the data support the view that the concentrations remain above the MIC at 96 h after the dose.

Estimate of the intradiscal antibacterial exposure with respect to the linezolid PK/PD driver of 24 h area under the curve (AUC)/MIC.

The murine neutropenic thigh model of bacterial infection has been used to estimate the pharmacodynamic parameters most predictive of the efficacy of linezolid. ⁴⁴ The model used plasma linezolid pharmacokinetics and bacterial burden of the thigh muscle under the assumption that plasma and thigh muscle linezolid concentrations are related. The concentration of muscle tissue linezolid in rats is 58%–80% of that in plasma during 24 h after a single intravenous administration. ⁴⁰ The linezolid PK/PD target for bacterial stasis, that is, no increase in bacterial burden, is a value for 24 h AUC/MIC of 10, and for substantial killing, a 3 log₁₀ reduction, an exposure which may sterilize tissue, is >1000. ⁴⁴

The mean disc AUC, the estimated amount of drug in disc, between 72 and 96 h after PP353 administration (AUC_72‐96h) was approximately 100 mg.h; therefore, the AUC_72‐96h would be 10 000 μg.h/mL, assuming a disc space of 10 mL. The 24 h AUC/MIC was 10 times the required linezolid exposure for a 3 log₁₀ reduction in bacterial burden.

The mean disc AUC between 96 and 120 h was estimated to be less than 30 μg.h/mL and therefore a 24 h AUC/MIC that may be consistent with bacterial stasis, but not killing. A second dose of PP353 at 96 h or within the 96–120‐h interval would be appropriate to maintain efficacious, bactericidal to bacteriostatic, exposure in the nucleus pulposus.

Selection of a PP353 dose regimen for Part B of the Persica 002 clinical trial.

The systemic pharmacokinetics of PP353 and the deconvolution analysis suggest that a single dose provides bactericidal exposure for up to 96 h and continued bacteriostatic exposure up to 120 h. Plasma linezolid concentration below 5 ng/mL is considered an indication that intradiscal linezolid has been substantially eliminated from the intervertebral disc space. The time post‐dosing to reach 5 ng/mL (T_dur) was between 89.4 and 110 h. As per the protocol, the Safety Review Committee (SRC) reviewed the PK data after the first three subjects had been dosed. The SRC considered the PK of PP353 was sufficiently consistent to allow dose selection to support 8 days exposure and progression to Part B of the study. The PK findings are consistent with the dose on Day 1 and Day 5 ± 1 in Part B of the Persica 002 study to provide approximately 8 days of antibacterial intradiscal linezolid exposure.

4. DISCUSSION

This first‐in‐human Phase 1b pharmacokinetic study of an intradiscal linezolid formulation, PP353, deployed a maximum feasible linezolid dose at 50 mg/mL and a maximum injection volume of 3 mL. The objective was to characterize single‐dose pharmacokinetics and then to assess whether one, two, or three doses of PP353 would be required to achieve 8 or more days of intradiscal exposure. The study demonstrated that a dose volume of 3 mL was practical and tolerated by subjects and that a single administration provided detectable plasma linezolid and, by extrapolation, intradiscal concentrations for 96–144 h. Deconvolution of the plasma concentration time curve allowed an estimate of the amount of linezolid released from the disc and by subtraction an estimate of the amount of linezolid remaining. An indirect estimate is necessary as it is not feasible to directly measure the concentration of linezolid in the intervertebral disc space. When plasma linezolid concentration fell to 5 ng/mL, the linezolid disc depot was essentially eliminated. The estimate for intradiscal linezolid exposure, compared to preclinical linezolid PK/PD studies, suggest that there may be bactericidal exposure for 96 h and bacteriostatic exposure for up to 120 h after dosing. The results were sufficiently consistent to allow dose selection after three subjects. One dose provided 4–5 days of exposure and hence a two‐dose regimen could provide 8 days of coverage.

Based on these findings a dose regimen for Part B of the study to explore safety, tolerability, pharmacokinetics, and efficacy of dosing with 3 mL of PP353 (150 mg linezolid) on Day 1 and Day 5 ± 1 was recommended.

Linezolid is commonly used to treat bacterial infection with an oral or intravenous dose of 600 mg twice a day (Q12h) for 10–14 days up to a maximum duration of 28 days. ³⁹ Major side effects such as myelosuppression are associated with prolonged exposure. In Part B, a two‐dose regimen will provide a total dose of 300 mg of linezolid over the 8 day treatment period compared to a daily recommended oral dose of 1.2 g and a standard course of treatment of 12–16.8 g. Two doses of 3 mL of PP353 is a substantially lower dose of linezolid than is typically used. Furthermore, the systemic C _max of linezolid when PP353 is administered is approximately 10% of that observed during standard iv or oral linezolid dosing. The low linezolid C _max and comparatively low dose of PP353 versus standard administration greatly reduces the likelihood of drug‐related adverse reactions. The study SRC has stated that drug interactions with SSRI and MAOI pose a lower risk with PP353 than with standard linezolid dosing and recommended that no medications should be contraindicated due to potential DDI in Part B of the study.

This study has limitations. The consistency of pharmacokinetic profiles of three subjects allowed selection of a dose regimen for Part B, but these subjects may not reflect a wider population. The deconvolution study and estimates of intradiscal exposure may also be biased due to the small sample size and lack of iv data in the subjects under investigation. The pharmacokinetics of a two‐dose regimen and deconvolution analysis will be confirmed in Part B of the study.

5. CONCLUSIONS

The first‐in‐human assessment of PP353, a linezolid formulation for intradiscal administration, in subjects with CLBP and MC1 or mixed MC1 and MC2 lesions, has provided evidence of tolerability of the dose volume and intradiscal administration procedure. Preliminary plasma‐time exposure pharmacokinetic analysis has allowed an estimate of intradiscal linezolid exposure and selection of a dose regimen for evaluation of the efficacy of PP353 to treat CLBP with MC1.

AUTHOR CONTRIBUTIONS

ST was the Study Chief Investigator. RS was a site Principal Investigator. AG and PD were co‐investigators. DM led the study protocol development and acted as the Sponsor's Chief Medical Officer. SG, LC, and SR led initiation and sponsor oversight of the trial. CB provided pharmacology expertise and contributed to study design. PC and SS led the manufacturing of drug product and oversight of pharmacy activities. CS led the operational management of the study. GS contributed to the study design. LC contributed to study design and provided the first draft of the manuscript. All authors edited and approved their sections and approved the manuscript in general.

FUNDING INFORMATION

Persica Pharmaceuticals Ltd. funded the study and was responsible for the conceptualization, overall design, collation of data, decision to publish and the first draft of the manuscript.

CONFLICT OF INTEREST STATEMENT

SG, SR, and LGC declare a financial interest and salary from Persica Pharmaceuticals Ltd., a company developing intradiscally administered antibiotics to treat chronic low back pain (CLBP) in clinical trials. Other authors work for research organizations contracted to provide expertise and services and their organizations received payment from Persica Pharmaceuticals Ltd. for the work detailed in this manuscript. ST, RS, AG, PD, DH, CB, PC, RD, SS, CS, and GS declare no financial interests relevant to the manuscript.

CONSENT

All subjects gave their consent. All subjects gave their consent. All authors agree with the content of the manuscript and its submission to the Journal.

CLINICAL TRIAL REGISTRATION

The study was registered at www.clinicaltrials.gov (reference NCT04238676).

Tripathi SS, Sneath R, Golash A, et al. Pharmacokinetics of PP353, a formulation of linezolid for intervertebral disc administration, in patients with chronic low back pain and Modic change Type 1: A first‐in‐human, Phase 1b, open‐label, single‐dose study. JOR Spine. 2024;7(4):e70009. doi: 10.1002/jsp2.70009

DATA AVAILABILITY STATEMENT

Persica Pharmaceuticals Ltd. is committed to provide access to anonymised data that underpin the results of clinical trials. Data from Persica‐sponsored clinical trials will be available for sharing once a medicine has been approved by regulators or terminated from development, and the study has been accepted for publication.