Peritonitis and Technique Failure Caused by Roseomonas mucosa in an Adolescent Infected with HIV on Continuous Cycling Peritoneal Dialysis

Abstract

We report the first case of peritonitis caused by Roseomonas mucosa which led to technique failure in an adolescent patient with HIV receiving peritoneal dialysis. Identification of the causative organism by 16S rRNA gene sequencing and phylogenetic analysis is described.

CASE REPORT

A 19-year-old male with end-stage renal disease on continuous cycling peritoneal dialysis (CCPD) presented with increasing abdominal pain for 4 days and 3 days of fever to 38.9°C. His medical history included chronic renal failure from IgM nephropathy/focal segmental glomerulosclerosis and perinatally acquired HIV with a viral load of <48 copies/ml and a CD4 cell count of 576/μl on therapy with ritonavir, atazanavir, abacavir, and lamivudine. He had developed renal failure 6 months before, and peritoneal dialysis (PD) was initiated after peritoneal catheter placement. Upon presentation to the emergency department, clinical peritonitis was diagnosed and dialysate fluid samples were obtained for culture, cell counting, and microscopy. Empirical treatment with intraperitoneal and intravenous ceftazidime and vancomycin was initiated. The dialysate fluid cell count showed 230 white blood cells per μl with 70% neutrophils (Table 1). The initial PD fluid culture was negative, but a second specimen obtained the following day grew a non-lactose-fermenting Gram-negative rod. The patient's symptoms resolved, and he completed a 14-day course of empirical intraperitoneal antibiotic therapy. At the end of therapy, analysis of dialysate fluid showed 1 white blood cell/μl but a dialysate culture that was initially negative became positive after 72 h and grew a non-lactose-fermenting Gram-negative rod. Five days after completing therapy, he returned to the emergency department with severe abdominal pain and analysis of dialysate showed a white cell count of 239/μl with 35% neutrophils (Table 1). Empirical treatment with intraperitoneal ciprofloxacin, gentamicin, and vancomycin was initiated because of perceived failure of the initial ceftazidime course. Five days after collection, the pretreatment dialysate culture grew a non-lactose-fermenting Gram-negative rod. Intraperitoneal ciprofloxacin was continued for a total of 21 days with resolution of symptoms, a normal dialysate fluid cell count, and negative culture at the end of therapy. A routine dialysate sample collected 1 week later grew Streptococcus agalactiae (group B Streptococcus) in the absence of symptoms of peritonitis or an elevated peritoneal fluid white cell count. He completed a 14-day course of intraperitoneal ceftriaxone, and as this was his third infection in 6 weeks, the peritoneal catheter was removed. Although the peritoneal catheter was later replaced, subsequent CCPD failed because of dense adhesions entrapping the new catheter.

Table 1.

Time line of infection and organism identification

Treatment course and day	PD fluid white blood cell count (cells/μl)/differential (%)a	Time to growth (days)	PD fluid culture result	Intraperitoneal treatment (duration [days])
1st
0	230/N, 70; L, 13; M, 15; E, 2		Negative	Ceftazidime + vancomycin (14)
2	830/N, 33; L, 6; M, 53; E, 4	4	Non-lactose-fermenting Gram-negative rod, unidentified
None, 15	1	3	Non-lactose-fermenting Gram-negative rod later identified as R. mucosa	None
2nd
20	239/N, 35; L, 5; M, 54; E, 6	5	Non-lactose-fermenting Gram-negative rod	Ciprofloxacin (21), gentamicin (7), vancomycin (5)
21	12/N, 21; L, 8; M, 67; E, 4		Negative
None, 43	16		Negative	None
3rd, 48	84/N, 0; L, 25; M75; E, 0	3	Group B Streptococcus	Ceftriaxone (14), PD catheter removal
None
62				PD catheter replacement
69				PD failed, long-term hemodialysis commenced

^aN, neutrophils; L, lymphocytes; M, monocytes; E, eosinophils.

Laboratory identification of bacterial isolates included standard and advanced diagnostics. At each sampling, 60 ml of peritoneal dialysate was centrifuged and the sediment was used for Gram staining and culture. This sediment was inoculated onto 5% sheep blood agar, MacConkey agar, and chocolate agar (BBL Microbiology Systems, Cockeysville, MD). After 72 h of incubation at 37°C, colonies on blood agar and chocolate agar were mucoid and pale pink but there was no growth on MacConkey agar. The Vitek II automated microbial identification system (bioMérieux, Marcy l'Etoile, France) first misidentified the organism as Ochrobactrum anthropi and later as Acinetobacter lwoffii, while the API 20 NE system (bioMérieux, Marcy l'Etoile, France) misidentified the causative organism as Methylobacterium mesophilicum; all are nonfermenting Gram-negative rods. However, given the phenotype of the organism and the Gram stain, a Centers for Disease Control and Prevention (CDC) pink coccoid species not identified by standard laboratory techniques was suspected. The organism was forwarded to the Microbiology Reference Laboratory at the Florida Department of Health Bureau of Laboratories (Jacksonville, FL), where it was identified as Roseomonas mucosa by biochemical testing (16) (Table 2).

Table 2.

Biochemical reactions of the patient's R. mucosa isolate^a

Characteristic or testb	Reaction
Motility	+
Morphology
Colony type	Mucoid
Pigment	Pink
Growth conditions
RT,c 35°C, 42°C	+
Salt broth 0%	+
Salt broth 6%	−
On cetrimide	−
On MacConkey slant	(+)
Enzyme activity
Ornithine decarboxylase	−
Arginine decarboxylase	−
Lysine decarboxylase	−
Oxidase	+
Catalase	+
Acid production from oxidative-fermentation sugars
Dextrose	−
Lactose	K
Maltose	K
Mannitol	−
Sucrose	K
Xylose	K
Other
Esculin hydrolysis	−
Gelatin liquefaction	−
Citrate	+
Indole production	−
Nitrate reduction	−
Phenylalanine	−
Sodium acetate	+
Hydrogen sulfide production	−
Urea utilization	+

^a+, positive; −, negative; K, alkaline; (+), weak growth.

^bAPI 20 NE profile, 0201044.

^cRT, room temperature.

Definitive identification of the organism was made by 16S rRNA gene sequencing and phylogenetic analysis. 16S primers fD1(5′-AGAGTTTGATCCTGGCTCAG-3′) and 536R (5′-CGTATTACCGCGGCTCGCT-3′) were added to 12.5 μl of a PCR master mix (Promega Corp., Madison, WI), 8.5 μl of H₂O, and 1.5 μl of boiled bacterial extract and thermocycled as follows: 1 cycle of 95°C for 5 min, followed by 35 cycles of 95°C for 30 s, 59°C for 30 s, and 72°C for 40 s and a final cycle of 72°C for 5 min. The universal primers described above amplify the first 536 bp of the 16S rRNA gene (according to the Escherichia coli numbering). Comparison of the sample 16S rRNA gene sequence with the total nucleotide collection in GenBank using the Basic Local Alignment Search Tool (BLAST) algorithm was used to assign the bacterial name with 99% similarity to other sequences. We also aligned several commonly occurring type strains of Roseomonas species with our strain. The assembled dendrogram was prepared by using a pairwise alignment tool and the unweighted-pair group method using average linkages cluster analysis algorithm in the BioNumerics v4.5 software package (Applied Maths, Austin, TX) (Fig. 1).

Fig 1.

Dendrogram of Roseomonas species based on the first 436 bp of the 16S rRNA gene. The American Type Culture Collection designations and GenBank accession numbers of the type strains are shown in parentheses.

Final identification and susceptibility testing of R. mucosa were completed months after the initial presentation, and the results were not available to guide treatment. The Etest (bioMérieux, Marcy l'Etoile, France) was performed in accordance with the Clinical and Laboratory Standards Institute interpretive criteria for non-Enterobacteriaceae (5). Mueller-Hinton agar plates (BBL Microbiology Systems) were inoculated with the equivalent of a 0.5 McFarland standard bacterial suspension, and MICs were read after 24 h. Pseudomonas aeruginosa ATCC 27853 and E. coli ATCC 25922 and ATCC 35218 were used as quality controls. The organism was susceptible to amikacin (MIC, 0.75 μg/ml), ceftriaxone (MIC, 0.75 μg/ml), ciprofloxacin (MIC, 0.125 μg/ml), levofloxacin (MIC, 0.125 μg/ml), gentamicin (MIC, 0.094 μg/ml), imipenem (MIC, 0.38 μg/ml), and meropenem (MIC, 0.125 μg/ml); intermediately susceptible to cefepime (MIC, 16 μg/ml); and resistant to piperacillin-tazobactam (MIC, >256 μg/ml), ceftazidime (MIC, 64 μg/ml), and trimethoprim-sulfamethoxazole (MIC, >32 μg/ml). Not surprisingly, the initial empirical antimicrobial regimen of ceftazidime and vancomycin for 14 days did not eradicate the organism. Although the eventual choice of ciprofloxacin and gentamicin for the second episode was fortuitous and appears to have cleared the infection, it was too late to prevent peritoneal scarring.

Infectious complications of PD can result in significant morbidity, hospitalizations, and death (19, 22). Recurrent peritonitis causes intra-abdominal inflammation and adhesions that may prohibit subsequent dialysis; it is the foremost reason patients change from PD to hemodialysis (26). Pathogenic organisms associated with peritonitis in the setting of PD include Gram-positive organisms such as Staphylococcus aureus, Staphylococcus epidermidis, Streptococcus species, and Gram-negative organisms such as E. coli, P. aeruginosa, and Klebsiella species, as well as anaerobes, fungi, and mycobacteria (3, 27). Episodes of culture-negative peritonitis can be vexing, as long courses of broad-spectrum antibiotics are often used to treat unknown organisms without known antimicrobial susceptibility profiles (11). Modifiable reasons for failure to identify a causative organism include improper sample handling and low-volume inoculation of culture media, but other factors that are more difficult to address, such as low organism burdens or atypical growth patterns, often cannot be determined in routine clinical practice (8). We report the first case of peritonitis due to R. mucosa, a slow-growing organism with low virulence that was challenging to identify and therefore treat, resulting in technique failure and subsequent hemodialysis.

Roseomonas is among a small group of organisms previously known as CDC pink coccoid groups I through IV (24). They possess a characteristic pink, mucoid-to-runny appearance on Sabouraud agar and are plump, Gram-negative rods or coccobacilli when observed upon Gram staining. They resemble M. mesophilicum and are commonly misidentified as such but can be differentiated by Gram stain morphology and the inability to oxidize methanol or utilize acetate (10, 16). Currently, there are five species/subspecies of Roseomonas that are known to be pathogenic in humans: R. gilardii subspecies gilardii, R. gilardii subspecies rosea, R. cervicalis, R. fauriae, and R. mucosa (10). In particular, R. mucosa has only recently been distinguished from R. gilardii (10, 16). Roseomonas species are more commonly implicated as etiologic agents in immunocompromised hosts with underlying illnesses, including chronic renal disease, diabetes, rheumatoid arthritis, malignancy, and AIDS (1, 4, 6, 7, 18, 23), but cases have been reported in the absence of immunocompromise (25). Of the five species/subspecies, R. mucosa is the one most commonly associated with clinical disease (6). Prior to this case report, only R. fauriae (2) and R. gilardii (17) were identified as Roseomonas subspecies/species associated with peritonitis in patients undergoing CCPD.

Roseomonas species are generally considered organisms with low pathogenic potential that can be successfully treated, but fatal disease has been described (6, 20, 21). As infection with Roseomonas species is uncommon, evidence describing the significance of infection is limited. Our case illustrates that repeated episodes of peritonitis due to R. mucosa may result in intra-abdominal adhesions causing mechanical restriction of the peritoneum and precluding PD. A modest elevation of the dialysate white cell count like that observed in our patient could be explained by the low virulence of Roseomonas species, which nonetheless induced inflammation and scarring of the peritoneal cavity.

The empirical antibiotic regimen for peritonitis in patients on PD usually includes a cephalosporin to cover Gram-negative, as well as Gram-positive, organisms (13). However, R. mucosa, in addition to being the most commonly reported species of the genus Roseomonas, is also estimated to have the greatest antibiotic resistance (10), with significant resistance to β-lactams. Roseomonas species are generally susceptible to aminoglycosides, fluoroquinolones, and carbapenems (1, 9, 10, 12, 14). A previous case series reported that 100% of R. mucosa isolates were resistant to ceftazidime but susceptible to amikacin and ciprofloxacin by the Etest, consistent with our results (10). Antimicrobial susceptibility testing of another isolate of R. mucosa found that, with the exception of cefepime and the carbapenems, the organism was resistant to all β-lactams by the Etest (4).

This patient's initial negative dialysate culture raises the possibility that some episodes of culture-negative peritonitis may, in fact, be due to fastidious Gram-negative organisms such as Roseomonas that are missed because of slow growth and low bacterial colony counts. Recently, a review of 435 episodes of culture-negative peritonitis found that patients who completed a treatment course with an aminoglycoside were less likely to experience catheter removal (10% versus 25%; P = 0.001) and permanent therapy transfer to hemodialysis (9% versus 17%; P = 0.06) than were patients whose aminoglycoside therapy was discontinued early (8). A recent report by Najafi et al. found that among patients in Iran with initial culture-negative peritonitis, there was a higher frequency of Gram-negative infections on subsequent positive dialysate cultures. They also found relatively low dialysate white cell counts in culture-negative peritonitis, which they attributed to a low organism burden. Proposed solutions to overcome this problem include multiple and large-volume samplings, as well as the use of rapid blood culture systems such as Bactec (BD Diagnostic Systems, Franklin Lakes, NJ), Septi-Chek (BBL Microbiology Systems, Cockeysville, MD), or BacT/Alert (bioMérieux, Marcy l'Etoile, France) for peritoneal culture in order to isolate the organism (13, 15).

In this case of recurrent peritonitis, the causative organism was identified by using advanced diagnostics and genotyping by a microbiology reference laboratory. Since R. mucosa has not previously been identified as an etiologic agent of peritonitis in continuous ambulatory peritoneal dialysis, it and other fastidious, slow-growing organisms should be considered when extending broad-spectrum antimicrobial therapy of culture-negative peritonitis. Such an approach may prevent peritoneal scarring, adhesions, and technique failure.