Skin ulcers and disseminated abscesses are characteristic of Serratia marcescens infection in older patients with chronic granulomatous disease

To the Editor:

Chronic Granulomatous Disease (CGD) is a primary immune deficiency resulting from mutations in any of four subunits of phagocytic cell NADPH oxidase. CGD patients are at risk for infections with catalase-positive bacteria and fungi [1-3]. Serratia marcescens infection of bone and soft tissue is a common presentation of CGD in infancy [4-6]. In older children and adults with CGD, infection with Staphylococcus aureus, Burkholderia cepacia and Nocardia species, or fungi such as Aspergillus make up the majority of infections [3,4]. Our experience indicates S. marcescens remains a significant cause of infection in older children and adults with CGD. These infections have a pattern different from that seen in CGD infants. Osteomyelitis is less common, skin infections form poorly healing ulcers [7], and infections frequently occur at multiple sites.

We describe 16 episodes of S. marcescens infections in 15 patients 8–39 years of age with CGD followed between 1999–2008 at the National Institutes of Health (Table 1). Only patients with culture documented S. marcescens were included in this report. All patients were participants in an IRB approved study of the natural history of CGD. Written informed consent specifically allowed study of infections and other complications of CGD.

Table 1.

CGD mutation, age at diagnosis of infection, clinical sites of infection and length of antibiotic therapy in patients presenting with S. marcescens infections

Patient number	CGD defect	Age in years at infection	Sites of infection#	Approx. Length of treatment (days)
1	gp91phox	30	Lung	36
2	p47phox	9	Lung	180
3	gp91phox	15	Lung	31
4	gp91phox	19	Deep calf and arm abscesses	97
5	p47phox	13	Lung	50
6	gp91phox	19	Arm abscess	14
7	gp91phox	8	Axillary lymph node abscess	14
8	gp91phox	27	Lung	14
9	gp91phox	20	Arm, leg and scrotal abscesses, epididymitis, prostatic abscess	180
10	p47phox	24	Face and arm abscesses	55
11A*	gp91phox	26	Lung	140
11B*	gp91phox	29	Arm abscess	55
12	gp91phox	20	Submental abscess, nares	14
13	gp91phox	20	Peripancreatic abscess and periaortic lymph node	104
14	gp91phox	24	Multiple paraspinal abscesses, gluteal abscess, hip osteomyelitis	90
15	gp91phox	39	Cervical lymph node	21

The gp91^phox refers to deficiency of this glycoprotein subunit of the phagocyte oxidase that is the cause of the X-linked form CGD.

The p47^phox refers to deficiency of this protein subunit of the phagocyte oxidase that is the cause of the most common autosomal recessive form of CGD.

phox: phagocyte oxidase

^*Patient 11 had two documented episodes of S. marcescens during the review period.

^#Note that patients 4, 7, 9, 10, 12, 13, and 14 had multiple sites of infection while the other patients had a single anatomic site of infection.

Three patients had additional documented S. marcescens infection at some time prior to the index period under review for this report. One of these three patients had a history of two previous S. marcescens lung infections. Each of these prior infections antedate the index review period by at least four years. We did not include these older episodes of infection, as detailed information about those infections was deemed less reliable.

During the review period, some patients had one or more infections characteristic of CGD with organisms other than S. marcescens. Only one (Patient #13, Table 1) developed two episodes of infection with S. marcescens. We conclude that they are distinct Serratia infections and not relapse of the first infection, because those two infections were three years apart. This assumption is consistent with a previous report noting that for most infections in CGD, re-infection is more common than persistent infection [8].

Sites of infection included lung parenchyma (6), vertebrae (1), lymph nodes (3), deep intra-abdominal or pelvic abscess (4), deep soft tissue abscess of limb (6), and superficial infection of skin (2). Seven patients were found to have metastatic infection at multiple sites (Table 1).

None of the 6 patients presenting with pneumonia had clinically apparent secondary sites of S. marcescens infection. Conversely, 7 of the 10 episodes of S. marcescens infection that did not involve pneumonia presented with widely metastatic dissemination. The characteristics of this dissemination included more than one non-contiguous site of skin abscess, deep muscle abscess in limbs and/or abscess in internal organs. A particularly informative example was Patient #9 (Table 1) who presented with skin and soft tissue abscesses (Figure 1). Epididymitis, apparent on physical examination, prompted pelvic Magnetic Resonance Imaging that revealed a prostate abscess (Figure 2).

Figures 1A-C.

Photograph of skin ulcers resulting from S. marcescens infection in Patient #9.

A. Shown in this image is an ulcerating lesion on the inner right thigh.

B. Shown in this image are ulcerating lesions on the left scrotum.

C. Shown in this image are lesions on the inner aspect of the right upper arm. Note that the upper lesion has ulcerated while at the time of this photograph the lower area of infection has not yet ulcerated, but also went on to ulcerate.

Figure 2.

Magnetic resonance image of the pelvis of patient #9. The two bright confluent round objects in the center of this image represent a large abscess in the prostate.

Mixed infection in the same biopsy sample (S. aureus or S. anginosis in addition to the Serratia) was identified in four patients, but in all cases the predominant organism by colony count appeared to be S. marcescens. Although the significance and relative impact of these additional organisms on the clinical course was unknown, antibiotic coverage was adjusted as necessary to include coverage for treatment of these additional organisms. In all cases, the S. marcescens isolate was sensitive to a broad range of antibiotic classes including carbapenems, quinolones, cephalosporins, aminoglycosides, extended-spectrum beta-lactams and trimethoprim-sulfamethoxazole. There was a tendency to include or use a quinolone as primary treatment in most cases. If a second antibiotic was used it tended to be either an intravenous carbapenem or intravenous high dose trimethoprim-sulfamethoxazole. In all of the cases these Serratia infections occurred in a setting where standard long term infection prophylaxis had included trimethoprim-sulfamethoxazole. Compliance with this long term prophylaxis was variable. Notably, 10 of 16 of the Serratia isolates were sensitive to this agent. The majority of patients received intravenous antibiotic therapy with two agents for at least a portion of the treatment period. One patient was treated with three intravenous antibiotics and one patient was treated with only one intravenous antibiotic (levofloxacin). Two patients received oral quinolone (ciprofloxacin or levofloxacin) therapy only (lymph node infection in one case and superficial abscess in the other; both without metastatic infection). Duration of therapy varied according to severity of the infection (range 14–180 days; mean 81 days; median 55 days). All infections were successfully eradicated; and all patients survived his or her infection.

Our current report represents the largest aggregate detailed description of Serratia infections in older children and adults with CGD. Previous reports describe either single patients or small case series [4-7], primarily in infants and young children. In other publications, the incidence of Serratia infection is reported incidental to an overall report of a registry of CGD patients without detailed description of the characteristics of the Serratia infections [3]. The important findings of our study reporting the details of 16 episodes of S. marcescens infections in 15 older children and adults with CGD are: 1. Serratia osteomyelitis was uncommon compared to that reported in infants with CGD; 2. Serratia infections of skin form large poorly healing ulcers; 3. Serratia pneumonia in our series was not associated with clinically apparent infection to extrapulmonary sites; 4. Seven of the ten (70%) non-pulmonary infections with Serratia in our series were associated with metastatic spread of infection to multiple sites. As clinically indicated by history, exam or laboratory findings, this group of patients often may require additional imaging studies (CT, MRI or PET scan) to demonstrate the presence of cryptic sites of infection. Finally, we note that all the infections in our series were eventually eradicated; indicating that full recovery from Serratia infection in CGD is an expected outcome with recognition of the variable presentation of this disease and with proper management.