Two Cases of Urinary Tract Infection Caused by Propionimicrobium lymphophilum

Abstract

The first case reports involving Propionimicrobium lymphophilum, a rarely encountered anaerobic Gram-positive non-spore-forming rod, are presented here as urinary tract infections. Initial detection of these bacteria required urine Gram stains. Comparison of the type strain to the two isolates by various methods is depicted and includes antimicrobial susceptibility data.

CASE REPORTS

C ase 1. A 79-year-old male presented to the primary care clinic with complaints of painful and frequent urination. The patient had a history of kidney and bladder stones, urinary tract infections, elevated prostate-specific antigen levels with negative prostate biopsy specimens, and an enlarged prostate accompanied by urinary retention. He was prescribed one tablet of sulfamethoxazole (800 mg)-trimethoprim (160 mg) (SXT) once daily for 10 days, an antibiotic that had alleviated an Escherichia coli urinary tract infection 8 months earlier. A clean catch urine specimen was submitted prior to starting the antibiotic. The urinalysis was positive for leukocyte esterase (LE), blood, and >300 mg/dl protein. Microscopic examination showed 15 to 20 white blood cells (WBC) per high-power-field (HPF), 10 to 15 red blood cells (RBC)/HPF, and abundant or 4+ bacteria using a scale of 0 to 4+. Normal ranges were as follows: <30 mg/dl protein, 0 to 5 WBC/HPF, and 0 to 5 RBC/HPF. A routine urine culture on 5% sheep blood agar (blood agar) (Thermo Fisher/Remel, Lenexa, KS) and MacConkey agar (BD, Sparks, MD) showed no growth after 48 h of incubation at 35°C under ambient air conditions. Upon completion of his antibiotic treatment, a follow-up urine sample was negative for LE and positive for blood and 100 mg/dl protein, with 0 to 2 WBC/HPF and 10 to 15 RBC/HPF; however, 4+ bacteria with a negative culture were still present. Symptoms had been relieved by the antibiotic. His physician requested three urine samples for cytological examination due to the hematuria.

Less than 2 weeks after completion of treatment, the increased urinary frequency with mild pain returned. The three urine cytology specimens were reported as free of tumor cells but exhibited abundant inflammatory cells and bacteria. Another clean catch specimen was obtained.

The patient's urine specimen was positive for LE, nitrite, blood, and protein with >50 WBC/HPF, 5 to 10 RBC/HPF, and 4+ bacteria. A Gram stain was performed on unspun urine which revealed many WBC and pleomorphic Gram-positive rods. The specimen was cultured using blood and MacConkey agars for 35°C aerobic incubation. Blood and chocolate agars (BD) were both incubated under 7% CO₂ and anaerobic conditions at 35°C. CDC anaerobe blood agar (CDC-ANA) (BD) was also used for anaerobic incubation. All plates were inoculated using a 1-μl loop. After a period of 72 h, growth was observed only on the anaerobic plates. The colonies were small (<1 mm in diameter), in numbers that indicated >10⁵ CFU/ml, and negative for catalase. A Gram stain showed Gram-positive rods with very pleomorphic morphology, including club-shaped and coccoid forms. The growth was subcultured to CDC-ANA for analysis via the RapID ANA II system (Thermo Fisher/Remel). Microcode 000671 resulted, indicating a 99.8% probability of Clostridium hastiforme with a probability level of “implicit.” The isolate lacked both colony characteristics and any microscopic resemblance to Clostridium species. A report of “>10⁵ CFU/ml anaerobic Gram-positive rod, non-sporeformer; resembles Actinomyces species” was released. The significance of finding a Gram-positive rod in urine was questioned. The patient was not treated at this time but was referred to urology.

At his urology appointment 2 weeks later, the patient indicated sharp pain upon urination with increased frequency. A renal and bladder ultrasound performed prior to the appointment failed to disclose any kidney or bladder stones. Another urine sample was collected to verify previous unusual culture results. The urologist prescribed ciprofloxacin (CIP) at 500 mg every 12 h for 15 days. The sample collected tested positive for nitrite, LE, and >300 mg/dl protein. Microscopic observation revealed >50 WBC/HPF, 10 to 15 RBC/HPF, and 4+ bacteria. An unspun urine Gram stain displayed the same pleomorphic Gram-positive rods seen previously. Three-day growth was subcultured for RapID ANA II analysis. The results duplicated the previous specimen: growth of >10⁵ CFU/ml anaerobic Gram-positive rods and a RapID ANA II microcode of 000671. Anaerobe susceptibility testing was not available at this time.

Symptoms resolved after treatment with CIP. Urinalyses were performed upon initial completion and at 2 and 4 months' posttreatment, exhibiting similar results. Urine dipstick testing was negative, with the exception of 30 mg/dl protein. The microscopic examinations confirmed 0 to 3 WBC/HPF and 0 to 5 RBC/HPF. Bacteria were reported as “none observed” in each case.

The final isolate in this case, designated NH-212, was sent to reference labs for identification: MIDI Labs, Inc., in Newark, DE, for partial 16S rRNA gene sequencing and the Culture Collection University of Gotëborg, Gotëborg, Sweden (CCUG). The CCUG reported an identification of Propionimicrobium lymphophilum after extensive phenotypic testing that included cellular fatty acid-fatty acid methyl ester (CFA-FAME) analysis and reactions from RID32S, ID32AN, and Coryne API test strips (bioMérieux, Marcy l'Etoile, France). The CCUG did not use the API strips to obtain identification from each strip separately. API test result reactions were combined with classical in-house tests and utilized their own huge database and software for identification. NH-212 was assigned the number CCUG 54740. The partial (approximately 500-bp) 16S gene sequencing at MIDI Labs was performed using universal primers at positions 0005F and 0531R. The region was PCR amplified and sequenced using DNA polymerase (Taq polymerase) and dye terminator chemistry. The samples were electrophoresed on an ABI Prism 3130xl genetic analyzer (Life Technologies, Carlsbad, CA). After a GenBank search with the Basic Local Alignment Search Tool (BLAST), the sequence identified a 99% match with “Propionibacterium lymphophilum” (GenBank accession no. AJ003056), now known as Propionimicrobium lymphophilum.

Case 2.

A 70-year-old extremely ill male was evaluated in urgent care. The patient had exhibited dramatic weight loss and a frequent, productive cough accompanied by brown sputum. He claimed to have not been evaluated by a doctor in many years. His urine sample was positive for LE and for nitrite and had 10 to 15 WBC/HPF and 4+ bacteria. An unspun urine Gram stain detected clumps of Gram-positive rods. Due to suprapubic tenderness and painful urination, the physician prescribed ciprofloxacin (CIP) at 250 mg twice daily for 7 days. The physician referred him for further evaluation for his ominous-looking chest X rays and recent weight loss. Urinary symptoms abated within 2 days after commencement of treatment. After 3 days, the original urine culture revealed growth only under anaerobic incubation conditions on blood agar and CDC-ANA. The growth of the small colonies of pleomorphic, Gram-positive rods indicated urine numbers of >10⁵ CFU/ml. The patient was lost to follow-up due to home hospice care admission. He died less than 2 months later of metastatic lung cancer.

This isolate was designated NH-1513. The same two reference laboratories were utilized as for the case 1 isolate. It was identified as Propionimicrobium lymphophilum by both MIDI Labs via partial 16S rRNA gene sequencing and by the CCUG using their phenotypic analysis methods (www.ccug.se). It was assigned the number CCUG 48580. Both methods of identification were as described for case 1.

Basic clinical laboratory procedures and anaerobic susceptibility testing were performed on the two patient isolates. The type strain (ATCC 27520) from the American Type Culture Collection (ATCC) in Manassas, VA, was obtained for comparison.

All three organisms grew equally well on blood agar, chocolate agar, phenylethyl alcohol blood agar (BD), Columbia CNA blood agar (BD), and CDC-ANA media. Anaerobic incubation for 4 days was performed as described in references 1, 2, and 3.

Figure 1A demonstrates the growth of NH-1513 at day 4, while Fig. 1B shows a Gram stain of an NH-1513 colony. Figure 1B closely resembles a previous illustration (4), photomicrograph (5), and description (3) showing pleomorphic, Gram-positive rods with coccoid and club-shaped forms.

FIG 1.

(A) Colonial appearance of isolate NH-1513 after 4 days of anaerobic incubation on a blood agar plate. The colonies were circular, light gray to white, and raised or convex with entire margins. NH-212 and ATCC 27520^T colonies appeared to be approximately 0.8 to 1.0 mm; NH-1513 was slightly larger at 1.2 mm. (B) Gram staining of isolate NH-1513 showing pleomorphism, including club-shaped and coccoid forms.

NH-1513 and ATCC 27520^T were assessed with the RapID ANA II system. Both yielded the same microcode as the case 1 isolate NH-212: 000671. Biochemical reactions are shown in Table 1. Using the RapID CB Plus (Thermo Fisher/Remel) with 6 h of incubation, analysis of NH-212 and ATCC 27520^T each produced the microcode 4024510. NH-1513 generated microcode 5024510. The only difference between the three organisms was the utilization of the glucose substrate by NH-1513. The RapID CB Plus is not intended for identification of anaerobic Gram-positive bacilli. In this case, the microcodes proved valuable for comparison of the clinical isolates and the type strain.

TABLE 1.

Biochemical profile of NH-212, NH-1513, and ATCC 27520^T using RapID ANA II microcode 000671

Reaction no.a	Reactive ingredient	Test code designation	Result
1	Urea	URE	−
2	ρ-Nitrophenyl-β,d-disaccharide	BLTS	−
3	ρ-Nitrophenyl-α,l-arabinoside	αARA	−
4	σ-Nitrophenyl-β,d-galactoside	ONPG	−
5	ρ-Nitrophenyl-α,d-glucoside	αGLU	−
6	ρ-Nitrophenyl-β,d-glucoside	βGLU	−
7	ρ-Nitrophenyl-α,d-galactoside	αGAL	−
8	ρ-Nitrophenyl-α,l-fucoside	αFUC	−
9	ρ-Nitrophenyl-n-acetyl-β,d-glucosaminide	NAG	−
10	ρ-Nitrophenylphosphate	PO4	−
11	Leucyl-glycine-β-naphthylamide	LGY	+
12	Glycine-β-naphthylamide	GLY	+
13	Proline-β-naphthylamide	PRO	+
14	Phenylalanine-β-naphthylamide	PAL	+
15	Arginine-β-naphthylamide	ARG	+
16	Serine-β-naphthylamide	SER	+
17	Pyrrolidonyl-β-naphthylamide	PYR	−
18	Tryptophan	IND	−

^aReactions 2 to 10 represent enzymatic hydrolysis of aryl-substituted glycoside or phosphoester. Reactions 11 to 17 represent enzymatic hydrolysis of arylamide substrate detected with RapID ANA II reagent. Reaction 18 represents utilization of the substrate detected by the RapID spot indole reagent.

The RapID ANA II does not include nitrate. The RapID CB Plus produced negative nitrite results. The urine samples from the two patients tested positive for nitrite. The RapID systems detect preformed enzymes. Another rapid enzyme test, the nitrate disk test (Key Scientific, Stamford, TX), was performed and also negative. Using the indole-nitrate broth method (Anaerobe Systems, Morgan Hill, CA), which requires growth, nitrite tested positive for both of the clinical isolates but negative for ATCC 27520^T. Nitrate reduction is a variable characteristic of Propionimicrobium lymphophilum (3, 6).

The antimicrobial susceptibilities were determined using the Etest method (bioMérieux) for benzylpenicillin, ceftriaxone, tetracycline, imipenem, amoxicillin-clavulanate, clindamycin, CIP, sulfamethoxazole-trimethoprim (SXT), and metronidazole. A suspension equal to a no. 1 McFarland standard according to the manufacturer's instructions for anaerobes was inoculated on brucella agar with 5% sheep blood, hemin, and vitamin K (Thermo Fisher/Remel). Clinical and Laboratory Standards Institute (CLSI) document M100-S24 under the anaerobe section was used for the antibiotic interpretive standards (7). The plates were incubated in an anaerobic atmosphere for 48 h and did not display difference at 72 h. For antibiotics whose breakpoints had not been established for anaerobes (vancomycin, SXT, and CIP), only MICs are shown in Table 2. The three test organisms were susceptible to most antibiotics tested but resistant to metronidazole and likely SXT at >32 μg/ml. It has been shown that brucella agar is not optimum for SXT anaerobic testing (8). SXT was also tested on horse blood agar (Hardy Diagnostics, Santa Maria, CA) to confirm results. Previous reports of P. lymphophilum susceptibility data were not found for comparison.

TABLE 2.

Antimicrobial susceptibilities of two isolates and the type strain of Propionimicrobium lymphophilum by the Etest method

Antimicrobial	MIC (μg/ml)a
	NH-212	NH-1513	ATCC 27520T
Benzylpenicillin	0.047 (S)	0.094 (S)	0.047 (S)
Amoxicillin-clavulanate	0.094 (S)	0.19 (S)	0.125 (S)
Ceftriaxone	0.50 (S)	1.5 (S)	0.38 (S)
Clindamycin	0.125 (S)	0.75 (S)	0.047 (S)
Tetracycline	1.0 (S)	0.50 (S)	1.0 (S)
Imipenem	0.47 (S)	0.125 (S)	0.094 (S)
Vancomycinb	0.38	0.50	0.38
Ciprofloxacinb	0.75	0.75	0.75
Metronidazole	>256 (R)	>256 (R)	>256 (R)
Trimethoprim-Sulfamethoxazoleb,c	>32	>32	>32

^aS, susceptible; R, resistant.

^bAnaerobic breakpoints have not been determined.

^cTested using both brucella and horse blood agars.

For confirmatory identification of this rare bacterium, the full 16S rRNA gene sequencing results from the two clinical isolates and the type strain were characterized at MIDI Labs using universal primers 0005F and 1513R. Samples were prepared and analyzed in the same manner as the partial sequencing and electrophoresed on an ABI Prism 3130 genetic analyzer (Life Technologies). Consensus sequences of the approximately 1,480-bp regions were compared to sequences in the Sherlock DNA Library MD16M2 version 2.22, which failed to produce a match. Sequence data for the three test organisms compared to the GenBank database by BLAST search matched each organism to Propionimicrobium lymphophilum (GenBank accession no. NR_114337) with 99% similarity. Samples NH-1513 and NH-212 had identical sequences with only a 2-base difference from ATCC 27520^T. The sequence alignment, or total difference, was 0.13%. To meet the requirements for species identification for Gram-positive anaerobes, as described in CLSI document MM18-A (9), the GenBank BLAST search was continued for both clinical isolates. This revealed 99% similarity to Propionimicrobium lymphophilum strain JCM 5829 and Propionimicrobium lymphophilum strain DSM 4903, the type strain (DSM 4903 = ATCC 27520). Brooklawnia cerclae strain BL34^T was the next closest match at 94% similarity. Brooklawnia spp. and Propionimicrobium spp. are members of the same family, Propionibacteriaceae. The next type strain encountered in the search was Propionibacterium propionicum DSM 43307, with 93% similarity. The ezTaxon database (http://www.ezbiocloud.net/eztaxon) for the sequences of the clinical isolates showed 99.9% identity to the type strain DSM 4903 for both isolates. The next closest match was Propionibacterium propionicum F0230a at 92.6% identity. The closest type strain was Granulicoccus phenolivorans DSM 17626 at 92.0% identity.

Propionimicrobium was proposed as a genus in the family Propionibacteriaceae in 2002, with a single species, Propionimicrobium lymphophilum (3). The authors described various name changes, including “Corynebacterium lymphophilum,” and “Propionibacterium lymphophilum.” The original work was accomplished in 1916 by Torrey, who described bacteria isolated from lymph nodes of patients with Hodgkin's disease, calling them “Bacillus lymphophilus” (5).

Diseases caused by Propionimicrobium lymphophilum are not well established. A literature search including all of its previous names did not reveal any published case reports. However, isolates from human sources are listed on the CCUG website (www.ccug.se; 2014) and include three from human blood. Other examples of isolate sources can be found in The Prokaryotes (10). The type strain ATCC 27520 originates from submaxillary tissue (www.atcc.org). Imirzalioglu et al. examined a large number of urine cultures by nucleic acid-based methods, directly detecting P. lymphophilum in an unspecified number of culture-negative, LE-positive urine specimens (11).

Successful treatment was achieved with CIP in both cases. Although CIP is not active against many anaerobes (12), the MICs here were only 0.75 μg/ml. One of the most common anaerobes isolated from human infections, Propionibacterium acnes, is well known for being susceptible to CIP (13, 14). Propionibacterium spp. belong to the same family as Propionimicrobium, the Propionibacteriaceae.

In conclusion, the first two case reports of P. lymphophilum infections have been presented here. With the increasing availability of new technologies, the frequency and spectrum of infections as well as the ecology of this bacterium may be clarified. Initial detection of P. lymphophilum in urine specimens will remain a problem due to the need for direct Gram staining of the specimen and nonstandard growth conditions.

ADDENDUM IN PROOF

For further characterization, the isolates NH-212 and NH-1513 were analyzed by matrix-assisted laser desorption ionization–time of flight mass spectrometry (MALDI-TOF MS) (Bruker Biotyper Microflex LT; Bruker Daltonics, Billerica, MA) at MIDI Labs, Newark, DE. The resulting spectra were compared to two validated libraries (Bruker 1/16/2013 and MIDI Labs 8/12/2013) using a proprietary algorithm to produce an identification score. The closest match for both isolates was Propionimicrobium lymphophilum, with scores of 1.873 (for NH-1513) and 1.954 (for NH-212) to the genus confidence level. P. lymphophilum ATCC 27520^T gave a score of 2.077.