Surgical Upper Extremity Infections in Immunosuppressed Patients: A Comparative Analysis With Diagnosis and Treatment Recommendations for Hand Surgeons

Abstract

Background: Immunosuppression is encountered in patients with oncologic, transplant, and autoimmune disorders. The purpose of this study is to provide guidance for physicians treating surgical hand and upper extremity (UE) infections in immunosuppressed (IS) patients. Methods: We retrospectively reviewed our database of patients presenting with UE infections over 3 years. IS patients were matched randomly to non-IS patients. Patient background, infection presentation, surgical evaluation, and microbiology variables were recorded. Infection variables included mechanism, location, and type. Outcomes included inpatient length of stay (LOS) and need for repeat drainage. Results: We identified 35 IS and 35 non-IS out of 409 UE infection patients. Patients most commonly had a hematologic malignancy (34%) as their IS class, and the most frequent immunosuppressive medication was glucocorticoids (57%). IS patients were more likely to be older and less likely to have a history of drug abuse or hepatitis C virus infections. IS infections were more likely to have idiopathic mechanisms, more likely to involve deeper anatomy such as joints, bone, tendon sheath, or muscle/fascia, and less likely to present with leukocytosis. IS cultures more commonly exhibited atypical Mycoplasma or fungus. There was no difference between IS and non-IS patients regarding LOS or recurrent drainage. Conclusions: Mechanism and white blood cell count are less reliable markers of infection severity in IS patients. Physicians treating infections in IS patients should maintain a higher suspicion for deeper involved anatomy and atypical microbiology. Nonetheless, with careful inpatient management and closer surveillance, outcomes in IS patients can approach that of non-IS patients.

Introduction

Infections in immunosuppressed (IS) patients are devastating and represent the second leading cause of death in this patient population.¹² Treating physicians are challenged by the absence of normal clinical cues traditionally used for timely diagnosis and management.⁸ These patients often present without a clear source of infection such as a history of trauma.¹¹ They can present with innocuous seeming signs and symptoms that can quickly progress to more dire outcomes. See case examples (Figures 1 and 2). In the IS population, easily treatable and common microorganisms (such as Staphylococcus and Streptococcus) may result in greater morbidity and mortality.¹⁴ For instance, Linder et al reviewed 148 patients with culture positive Streptococcus pyogenes infections and found higher rate of necrotizing fasciitis and septic shock in the IS subset.¹⁰ Thus, a thorough understanding of infections in this patient population is necessary.

Figure 1.

A 46-year-old right-hand-dominant man who presented with a “paronychia” of the index finger.

Note. (a) He underwent incision and drainage in the emergency department and was discharged on oral antibiotics. He had discontinued Enbrel on the day of presentation (under the direction of his rheumatologist), and this tumor necrosis factor α blocker was not listed in his routine home medications. (b) He returned to the hospital with worsening pain, and further questioning elicited a history of human bite when breaking up an altercation. He required serial surgical intervention for a septic distal interphalangeal joint and flexor tenosynovitis. (c) He ultimately required partial amputation, secondary to the extent of the infection, weeks of wound care, and intravenous antibiotics.

Figure 2.

A 24-year-old right-hand-dominant woman presented with bilateral wrist pain and history of lupus, currently on prednisone.

Note. (a) She initially had a previous right radiocarpal arthrocentesis resulting in a positive Escherichia coli culture. She was subsequently taken to the operating room for right wrist incision and drainage (I&D) and left radiocarpal arthrocentesis. The right wrist intraoperative cultures were negative and she was then continued on intravenous antibiotics at home. (b) Initial right wrist radiograph. (c) Initial left wrist radiograph. (d) Re-presentation with worsening right wrist pain, joint collapse, requiring return to the operating room for left wrist I&D.

The IS are also at risk of developing clinically relevant infection due to atypical and opportunistic pathogens.^2,22 Specific disease processes as well as immunosuppressive medications change the effectiveness of the immune system. Therefore, the pathogenic microorganisms may vary significantly in, for example, those patients on treatment for inflammatory bowel disease versus those with AIDS.^15,20 Awareness of these differences may be helpful in guiding empiric antibiotic therapy and obtaining appropriate cultures.

Previous publications regarding hand and upper extremity (UE) infection in IS patients have focused on individual microorganisms or specific patient populations (inpatients, operative patients, etc).^3,5,13,23 Past work has not provided hand surgeons the framework to make complex and expedient decisions about a diverse population of patients that may need operative treatment and/or inpatient admission.

Therefore, this purpose of our work is to share data from our database of patients with surgical UE infections. We performed a comparative study of IS versus non-IS patients evaluated for UE infections by our division to: (1) examine how IS alters the distribution of disease with regards to presentation, microbiology, and outcomes; and (2) provide diagnostic and evaluation recommendations for physicians encountering these complex patients.

Materials and Methods

Institutional review board approval was obtained. We created a database of all surgical UE infection patients over 18 years of age who were evaluated by our department over 36 months from July 2014 to July 2017. This included patients presenting in the emergency department and inpatient wards of our tertiary care academic center (level I trauma center, National Cancer Institute accredited cancer center, major transplantation center that also provides indigent care for the major metropolitan area and outlying >5 state region). “Upper extremity” was defined as occurring in the forearm, hand/wrist, digits, or thumb. “Surgical infection” was defined as anyone from whom tissue or fluid cultures were collected and subsequent antibiotic treatment was initiated. The decision to obtain cultures and place the patient on antibiotics was at the discretion of the consulting physician based on the presenting clinical scenario.

For this study, we identified the subset of patients who were actively IS at time of consultation. IS was defined as patients who were on immunosuppressive agents currently or within 3 months,²⁴ or anyone with known AIDS (CD4 count less than 200cells/mm³). Exclusion criteria included nonsurgical infections that did not require drainage (ie, cellulitis) and patients presenting with presumed septic arthritis that either never required operative intervention or had positive crystal analysis with negative culture results.

IS diagnoses were divided into the following categories: solid organ transplant, autoimmune disease, hematologic malignancy, or metastatic malignancy (on chemotherapy). Active immunosuppressive agents were grouped based on broad mechanisms of action: glucocorticoids, tumor necrosis factor α (TNF-α) inhibitors, other biologics, antimetabolites, alkylating agents, topoisomerase inhibitors, mitotic spindle inhibitors, calcineurin inhibitors, and other.

Controls were selected from the non-IS group using a random number generator and 1:1 matching. Patient variables collected included age, sex, hand dominance, and the presence of comorbidities (diabetes, obesity, history of intravenous drug abuse [IVDA], history of hepatitis C virus [HCV] infection, and history of previous surgical UE infection), initial white blood cell (WBC) count, antibiotics at time of consult, inpatient status at time of consult, and presence of crystalline arthropathy.

Infection variables collected included type, etiology, and location. Type was classified as paronychia, felon, abscess, tenosynovitis, osteomyelitis, joint, and necrotizing fasciitis. Etiology was categorized as burn, fight bite, animal bite, IVDA, other trauma, or nontraumatic. Location was thumb, digit, hand/wrist, or forearm. Information on laterality was also collected. Multiple locations and/or simultaneous types of infection were all recorded for analysis. Microbiology laboratory data were categorized by the number of organisms, Gram stain morphology, mycobacterial, and/or fungal stain morphology and species of infectious agent. Treatment path data were subdivided and categorized as follows: (1) bedside incision and drainage (I&D) and discharge; (2) bedside I&D and admission for intravenous (IV) antibiotics; or (3) operative I&D and admission for IV antibiotics. The main clinical outcomes were inpatient length of stay (LOS) and surgical recurrence (defined as need for repeat drainage).

Summary statistics were calculated. Wilcoxon rank sum and chi-square analysis was used to compare continuous and categorical variables of interest, respectively. Alpha < 0.05 indicated statistical significance in all tests.

Results

Out of a total of 409 patients presenting with a surgical UE infection, 70 patients were included in this study: 35 IS versus 35 non-IS. Comparing IS with non-IS patients, IS patients were older (mean 53 vs 46 years, P = .043) and less likely to have a history of IVDA (3% vs 31%, P = .002) or HCV infection (3% vs 29%, P = .003). There were no statistically significant differences with regard to sex, diabetes, obesity, or previous history of UE infection. In the IS group, the most common immunosuppressive disease category was hematologic malignancy (on chemotherapy) (34%), followed by autoimmune disease (31%) and solid organ transplant (11%) (Table 1). The most frequent immunosuppressive medication was glucocorticoids (57%), followed by antimetabolites (23%) and other biologics (14%) (Supplemental Table S1).

Table 1.

Patient Background Characteristics.

	Immunosuppressed cases (n = 35) n (%)	Nonimmunosuppressed controls (n = 35) n (%)	P value
Age (years)a	53 [46-70]	46 [33-58]	.043
Sex	Male 25 Female 10	Male 22 Female 13	.445
Diabetes	12	8	.290
Obesity	8	14	.122
Hypertension	14	10	.314
History of IV drug abuse	1	11	.002
HIV	2	0	.151
HCV	1	10	.003
Previous history of surgical hand infection	7		.490
Immunosuppression diagnosis class	Solid organ transplantb 4 Autoimmune diseasec 11 Hematologic malignancyd 12 Othere 8	N/A	—
Active immunosuppression medication class	Glucocorticoids 20 Antimetabolites 8 Alkylating agents 3 Topoisomerase inhibitors 3 Mitotic spindle inhibitors 2 Biologics 5 Calcineurin Inhibitors 4 TNF-α inhibitors 2 Other 4	N/A	—

Note. Controls were matched to cases based on age and diabetes history. IV = intravenous; HCV = hepatitis C virus; TNF-α = tumor necrosis factor α.

^aMedian [interquartile range].

^bSolid organ transplant included heart (2) and kidney (2).

^cAutoimmune disease included rheumatoid arthritis (3), lupus (3), inflammatory bowel disease (2), temporal arteritis (1), multiple sclerosis (1), and pemphigus vulgaris (1).

^dHematologic malignancy included B-cell lymphoma (1), Hodgkin lymphoma (1), acute myeloid leukemia (5), myelodysplasia (2), chronic lymphocytic leukemia (1), chronic myeloid leukemia (1), and Burkett lymphoma (1).

^eOther included eczema (1), pancytopenia (1), HIV/AIDS (2), glomerulonephritis (1), and metastatic malignancy (3).

The etiology of infection was more likely to be nontraumatic in the IS patient population compared with the non-IS group (60% vs 26%, P < .001). IS infections were more likely to present as septic arthritis (26% vs 9%), necrotizing fasciitis (9% vs 0%), or multiple types simultaneously (14% vs 9%) (P = .048). Among IS joint infections, the most common location was in the wrist (8/12, 75%). IS patients were less likely present with leukocytosis than non-IS patients (31% vs 57%, P = .021). Regarding clinical outcomes, there was no difference between IS and non-IS patients in terms of LOS (median 5 vs 4 nights, P = .127) or surgical recurrence (34% vs 37%, P = .803) (Table 2, Figure 3).

Table 2.

Clinical Infection and Surgical Characteristics.

	Immunosuppressed patients (n = 35) n (%)	Nonimmunosuppressed controls (n = 35) n (%)	P value
Mechanism	Burn 0 Fight bite 1 Animal bite 3 IV drug abuse 0 Other trauma 10 Idiopathic 21	Burn 1 Fight bite 1 Animal bite 4 IV drug abuse 7 Other trauma 13 Idiopathic 9	<.001
Location	Isolated  Digit 12  Thumb 4  Hand/wrist 13  Forearm 1 Multiple 5  Hand/wrist and forearm 4  Digit, hand/wrist, and forearm 1	Isolated  Digit 12  Thumb 4  Hand/wrist 12  Forearm 6 Multiple  Digit and hand/wrist 1	.048
Type	Isolated  Paronychia 3  Felon 0  Abscess 12  Tenosynovitis 2  Osteomyelitis 1  Joint 9  Necrotizing fasciitis 3 Multiple 5  Osteomyelitis & joint 2  Abscess & joint 1  Paronychia, osteomyelitis, & tenosynovitis 1  Felon & osteomyelitis 1	Isolated  Paronychia 2  Felon 4  Abscess 21  Tenosynovitis 1  Osteomyelitis 1  Joint 3  Necrotizing fasciitis 0 Multiple 3  Felon & joint 1  Paronychia, abscess, tenosynovitis, & osteo 1  Abscess & joint 1	.048
Joint location	DIP 1 PIP 0 MCP 2 CMC 1 Wrist 8	DIP 2 PIP 0 MCP 1 CMC 0 Wrist 1	.196
WBCa,b Low [⩽3.8] Normal [3.8-9.9] Leukocytosis [⩾9.9]	7.6 [3.9–12.6] 9 14 11	10.1 [8.5-12.4] 0 15 20	.021
Preconsultation antibiotics	24	18	.223
Preconsultation status	Outpatient 21 Inpatient 14	Outpatient 26 Inpatient 9	.309
Surgical triagec	Bedside I&D, home 2 Bedside I&D, admit 17 OR I&D, admit 16	Bedside I&D, home 7 Bedside I&D, admit 15 OR I&D, admit 13	.044
Clinical outcomes	LOSb,d 5 [3-7] Surgical recurrencee 12	LOSb,d 4 [1-6] Surgical recurrence 13e	.127 .803

Note. IV = intravenous; DIP = distal interphalangeal joint; PIP = proximal interphalangeal joint; MCP = metacarpophalangeal joint; CMC = carpometacarpal joint; WBC = white blood cell; I&D = incision and drainage; OR = operating room; LOS = length of stay. Bold denotes significance.

^aWBC count.

^bMedian [interquartile range].

^cClassified by initial surgical treatment: bedside drainage and disposition on oral antibiotics; bedside drainage and admission for intravenous antibiotics; operative drainage and admission for intravenous antibiotics.

^dLOS measured in inpatient nights.

^eDefined as need for repeat surgical drainage during treatment course.

Figure 3.

Kaplan-Meier curve showing inpatient length of stay.

Note. There was no statistically significant difference between IS and non-IS inpatients treated for surgical upper extremity infections, suggesting that outcomes can be similar between the two groups with a thoughtful and aggressive approach toward IS patients. IS = immunosuppressed.

Regarding microbiology, there was no difference in terms of number of organisms on final cultures (P = .490). IS cultures exhibited a higher but not statistically significant rate of atypical Mycoplasma (3% vs 0%, P = .321) and fungal (9% vs 3%, P = .317) organisms, and a lower yet not statistically significant rate of overall Staphylococcus aureus (31% vs 43%, P = .348) infection (Table 3).

Table 3.

Pathogenic Microbiology.

	Immunosuppressed patients (n = 35) n (%)	Nonimmunosuppressed controls (n = 35) n (%)	P value
Number of organisms	None 12 Mono-microbial 17 Poly-microbial 6	None 8 Mono-microbial 18 Poly-microbial 9	.490
Gram-stain morphology	Gram positive  MSSA 6  MRSA 5  Other 10 Gram negative 3 Mycoplasma 1 Fungal 3	Gram positive  MSSA 9  MRSA 6  Other 10 Gram negative 6 Mycoplasma 0 Fungal 1	.348 .703 .733 .263 .321 .317
Prevalence of specific microbiology, by agent	Glucocorticoid (n = 18)  No growth 8  Medicopsis romeroi 1  Coagulase-negative staphylococci 1  MSSA 3  Streptococcus dysgalactiae 1  MRSA 2  Escherichia coli 1  Bifidobacterium 1  Staphylococcus warneri 1  Mycoplasma 1 Antimetabolite (n = 10)  No growth 5  Medicopsis romeroi 1  Coagulase-negative staphylococci 2  Actinomyces 1  Abundant mixed 1  E coli 1  Bifidobacterium 1  Fusarium 1  MSSA 1 Alkylating agent (n = 3)  S warneri 1  Mycoplasma 1  MSSA 1 Topoisomerase inhibitor (n = 3)  S warneri 1  No growth 1  Mycoplasma 1 Mitotic spindle inhibitor (n = 2)  S warneri 1  Mycoplasma 1 Biologic (n = 5)  S warneri 1  No growth 1  Mycoplasma 1  MSSA 2 Calcineurin inhibitor (n = 8)  No growth 7  MRSA 1 TNF-α inhibitor (n = 2)  Actinomyces 1  Coagulase-negative staphylococci 1  Staph Staphylococcus intermedius 1  S intermedius 1  Streptococcus mitis 1  Streptococcus constellatus 1 Other (n = 3)  MSSA 2  No growth 1	N/A	—

Note. MSSA = methicillin-sensitive Staphylococcus aureus; MRSA = methicillin-resistant Staphylococcus aureus; TNF-α = tumor necrosis factor α.

Discussion

Surgical infections of the UE are commonly evaluated by hand surgeons. However, IS can enhance the complexity of this pathology, and these patients can have severe sequelae if not managed appropriately initially (Figure 1 and 2). Therefore, the purpose of this study was to review our experience treating IS versus non-IS surgical UE infections to provide preliminary recommendations for initial evaluation and treatment (Figure 4).

Figure 4.

Proposed evaluation algorithm.

^aSend fluid as well as tissue cultures if possible.

^bIf bony involvement, consider sending a proximal bony margin to pathology to assess for evidence of osteomyelitis, as this may help guide duration of antibiotic therapy.

^cConsider sending atypical fungal and Mycobacterial cultures for non-immunosuppressed patients with unusual presentations, such as acute or chronic infections, or freshwater exposure.

We observed noticeable clinical differences between IS and non-IS patients with regard to mechanism, presentation, and microbiology, with important implications. First, IS patients tended to be older, but not necessarily more comorbid with regard to diabetes and obesity. Second, IS patients were less likely to present with leukocytosis. Therefore, physicians should be cautioned that WBC count constitutes a less reliable marker of infection severity in this patient population, where a normal WBC does not necessarily rule out severe disease.

Third, IS patients more frequently presented with an idiopathic etiology. In our series of 35 IS patients, 60% had nontraumatic etiologies, compared with 26% of non-IS patients. In comparison, Fowler et al found only 2% (30 patients) with nontraumatic etiologies in a series of 1507 non-IS patients presenting with UE infection.³ Therefore, the mechanism itself is a less reliable guide to infection type and severity in this patient population as well.

Fourth, IS patients were more likely to present with clinically severe infections that involved deeper anatomy, despite absence of leukocytosis or antecedent trauma. Among non-IS patients, the prevalence of superficial anatomic involvement (paronychia, felon, abscess) was 89%, while the prevalence of deep anatomic involvement (tendon sheath, bone, joint, muscle/fascia) was 26%. In contrast, among IS cases, these proportions were 51% and 69%, respectively. These findings reflect the compromised ability of the immune system to fight the spread of infections in IS patients, which has been observed in the general literature. For instance, Linder et al showed in patients with a positive blood culture for S pyogenes, there were significantly more necrotizing infections in their IS group compared with the non-IS group (P < .001).¹⁰ Kaandrop et al observed rheumatoid arthritis to be an independent risk factor for joint infections,⁶ while Saraux et al observed HIV to constitute an independent predictor of septic arthritis.¹⁸ Hence, physicians evaluating surgical UE infections in IS patients should maintain an increased suspicion for deeper anatomic involvement.

Our study did not observe a statistically significant difference in distribution of microbiology between IS versus non-IS patients. However, we did observe a trend toward a higher frequency of atypical fungal or Mycobacterial infections in IS patients (11.4%) than non-IS patients (2.8%), though this did not reach significance (P = .320). Unfortunately, our study was likely underpowered to detect specific differences in microbiology, due to the low numbers and rare subpopulation of IS patients.

Nonetheless, atypical infections have been shown to occur more commonly in IS patients,^2,22 due to the various immunosuppressive agents which broadly decrease the immune response.^{1,9,16,21,25,26} For instance, glucocorticoids act through various molecular pathways, including mitogen-activated protein kinases to induce transcription of anti-inflammatory proteins and inhibit proinflammatory proteins.¹⁶ TNF-α inhibitors, which are used for treatment of inflammatory arthritis and other autoimmune diseases, prevent macrophage activation, recruitment, and granuloma formation.¹⁹ Patients on this agent remain at increased risk of granulomatous infections⁴ and fungal infections, the latter of which prompted a Food and Drug Administration “black box warning.”¹⁷ Therefore, we do recommend empirically obtaining atypical fungal and Mycobacterial stains and cultures on IS patients. These tests (despite their costliness) should also be obtained even in the setting of observed purulence, as Kazmers et al have shown 86% of their atypical cultures also had purulence at the time of drainage.⁷

Last, and importantly, we did not observe a statistically significant or clinically meaningful difference between IS and non-IS patients in terms of need for repeat drainage or inpatient LOS. This suggests that with a thoughtful approach, closer surveillance of IS patients, and increased suspicion that IS patients may present with deeper infected anatomy and possible higher risk of atypical microbiology, number of washouts and LOS can nonetheless approach that of non-IS patients.

Our study has several limitations. First, despite 3 years of enrollment at a high-volume tertiary center, we obtained a small number of IS patients, reflecting the rare subpopulation of this group. Second, our definition of IS included patients who were on immunosuppressive medications within 3 months of consultation. As each immunosuppressive medication has varying half-lives and durations of action, choosing a single time point for inclusion would not be feasible. However, there is evidence that IS can have lasting effects beyond medication cessation. Verma et al showed depleted levels of B cells and T cells for 3 to 9 months after cessation of chemotherapy for breast cancer patients.²⁴ Therefore, we felt a 3-month cutoff to be a reasonable time frame for this study. Third, this is a single institution study and these results may not necessarily be generalizable to other institutions. Fourth, a majority of patients in our study were on antibiotics prior to consultation. In addition, there may be inherent bias in IS patients to have earlier arthrocentesis or antibiotic use and this could increase the rate of false positive patients in our series. Fifth, our study specifically addresses infections which required surgical intervention and does not provide information on infections treated nonsurgically such as cellulitis. At our institution, these patients are managed by medical services, and we cannot provide information that addresses this larger population. Sixth, other variables—such as occupation, smoking status, occupational exposure—could certainly influence the microbiology and subsequent disease course of these patients, but these were not collected.

In conclusion, physicians evaluating surgical UE infections in IS patients should maintain an increased suspicion of deeper anatomic involvement and rely less on mechanism and WBC count as markers of infection severity. In addition, physicians should consider sending additional Mycobacterial and fungal cultures in this subpopulation, in addition to routine Gram stain and aerobic/anaerobic cultures. Empiric antibiotic coverage should be made in conjunction with formal infectious disease expert consultation. Future research is required to optimize the management plan for UE infections in IS patients and examine eventual functional outcomes in this patient population.