Outcome analysis in patients with primary necrotizing fasciitis of the male genitalia

ADULT UROLOGY

OUTCOME ANALYSIS IN PATIENTS WITH PRIMARY NECROTIZING FASCIITIS OF THE MALE GENITALIA PHILIPP DAHM, FRANK H. ROLAND, STEVEN N. VASLEF, RICHARD E. MOON, DAVID T. PRICE, GREGORY S. GEORGIADE, AND JOHANNES VIEWEG

ABSTRACT Objectives. To characterize patients with primary necrotizing fasciitis of the male genitalia (Fournier’s gangrene) and to identify risk factors and prognostic variables of survival. Methods. Fifty consecutive patients with primary necrotizing fasciitis of the male genitalia treated at our institution during a 15-year period between 1984 and 1998 were retrospectively analyzed. Of these patients, 44 (88.0%) were found to be eligible for analysis of the outcome parameters. Univariate survival analysis was performed using the Kaplan-Meier algorithm followed by multivariate analysis of statistically significant variables. Six patients (12.0%) who were severely immunocompromised were studied separately. Results. Medical comorbidities were prevalent, with diabetes being the most common condition (50%). The overall mortality rate was 20% (10 of 50). Three statistically significant predictors of outcome were identified among the variables analyzed. These were the extent of the infection (P ⫽ 0.0262), the depth of the necrotizing infection (P ⫽ 0.0107), and treatment with hyperbaric oxygen (P ⫽ 0.0115). Multivariate regression analysis of these variables identified the extent of the infection (P ⫽ 0.0234) as the only statistically significant, independent predictor of outcome in the presence of other covariables. Conclusions. The involved body surface area appears to be the most important prognostic variable, with a significant impact on outcome. Given the high mortality of the disease entity and a trend toward the improved survival of patients receiving hyperbaric oxygen, this treatment form appears indicated in more severe cases. Immunocompromised patients, who frequently have an atypical and fulminant clinical course, appear to constitute a separate group with a dismal prognosis. UROLOGY 56: 31–36, 2000. © 2000, Elsevier Science Inc.

F

ournier’s gangrene is a fulminant, necrotizing infection that originates from the perineal and genital area and rapidly extends along fascial planes to involve the groins, thighs, and abdominal wall. It has remained a highly lethal disorder1 since its recognition as a disease entity by Jean Alfred Fournier in 1883.2 The variables that influence the outcome of patients with Fournier’s gangrene remains controversial in large part because many series have been small and the statistical analysis remains limited. Major controversy has recently focused on the impact of the involved body surface area (BSA)3 and the indication to treat these patients with hyperFrom the Department of Surgery, Division of Urology and Division of General Surgery, and Department of Anesthesiology, Duke University Medical Center, Durham, North Carolina Reprint requests: Johannes Vieweg, M.D., Division of Urology, Duke University Medical Center, Box 2621, Durham, NC 27710 Submitted: August 31, 1999, accepted (with revisions): February 28, 2000 © 2000, ELSEVIER SCIENCE INC. ALL RIGHTS RESERVED

baric oxygen (HBO)4,5 in an attempt to improve tissue oxygenation. Small case series have further drawn attention to the poor outcome of severely immunocompromised patients with Fournier’s gangrene.6 – 8 In this study, we investigated these issues using multivariate statistical analysis of a large series of 50 consecutive patients with Fournier’s gangrene treated at a single institution during a 15-year period. MATERIAL AND METHODS The medical records of 50 consecutive patients admitted to our institution during a 15-year period between 1984 and 1998 were retrospectively reviewed. The diagnosis of primary necrotizing fasciitis of the male genitalia (Fournier’s gangrene) was established clinically on the basis of the patient’s history and physical examination and by radiologic imaging in select cases. The prognostic variables analyzed were the patient’s age, history of diabetes or chronic alcoholism, white blood cell count on admission, results of blood cultures, source of infection, and extent and depth of the infection. Additional parameters were the interval between the onset of 0090-4295/00/$20.00 PII S0090-4295(00)00604-X 31

TABLE I. Univariate and multivariate analysis of outcome variables of patients with Fournier’s gangrene (n ⴝ 44) Parameter Extent Small (0.5%–3% BSA) Medium (3.5%– 5% BSA) Large (⬎5% BSA) HBO Yes No Depth of necrosis Superficial Deep Blood cultures Negative Positive Source of infection Genitourinary Colorectal Unknown Chronic alcoholism No Yes Diabetes No Yes Colostomy No Yes Cystostomy No Yes WBC on admission Elevated Normal Age (yr) ⱕ60 ⬎60 Time to first debridement (days) 0–4 ⬎4

Patients

Mortality (n)

Median Survival

24

0 (0.0)

—

9

1 (11.1)

45.0

11

4 (36.4)

41.7

38 6

3 (7.9) 2 (33.3)

67.3 48.5

31 13

1 (3.2) 4 (30.8)

74.4 52.1

33 8

2 (6.1) 1 (12.5)

73.6 44.0

19 20 5

2 (10.5) 2 (10.0) 1 (20.0)

44.6 67.2 54.0

33 11

2 (6.1) 3 (27.3)

68.8 55.5

21 23

4 (19.0) 1 (4.3)

64.5 54.0

26 18

3 (11.5) 2 (11.1)

68.6 63.2

33 11

4 (12.1) 1 (9.1)

64.7 69.4

28 10

2 (7.1) 2 (20.0)

53.8 60.1

25 19

4 (16.0) 1 (5.3)

42.9 72.0

25 19

3 (12.0) 2 (10.5)

P Value Univariate

Multivariate

0.0262

0.0234

0.0115

0.0677 (NS)

0.0107

0.3918 (NS)

0.3733 (NS)

—

0.6784 (NS)

—

0.0538 (NS)

—

0.2859 (NS)

—

0.8727 (NS)

—

0.9253 (NS)

—

0.3594 (NS)

—

0.2788 (NS)

—

0.8945 (NS)

—

67.9 51.8

KEY: BSA ⫽ body surface area; HBO ⫽ hyperbaric oxygen; NS ⫽ not significant; WBC ⫽ white blood (cell) count. Numbers in parentheses are percentages.

clinical symptoms and the initial debridement and auxiliary measures such as HBO treatment and urinary and fecal diversion (Table I). The BSA involved by the infection was assessed using operative reports and Polaroid photo documentation and was quantified using the burn index9 at the time of maximum extent during the disease course. The involved BSA ranged from 1.0% to 23% (median 4.0%). According to the observed normal distribution of the BSA, the extent was stratified into small (BSA 0.5% to 3%), medium (BSA 3.5% to 5%), and large (greater than 5.0%), representing a localized, medium, and widespread disease process, respectively. The depth of the necrotizing infection was categorized as 32

either superficial, with involvement of the skin and subcutaneous tissue, or deep, with additional focal musculofascial, perivesical, or retroperitoneal involvement. The source of infection was distinguished as originating from a complicated urinary tract infection, a colorectal source, or of unknown etiology. Patients with a simple scrotal or perirectal abscess without necrotizing infection were not included in this series. Patients were promptly evaluated for HBO treatment either in the emergency department or immediately after surgery in the surgical intensive care unit. Indication for HBO treatment consisted of clinical evidence of a necrotizing mixed aerobic/ anaerobic infection. Patients with contraindications such as bullous lung disease or a compromised pulmonary gas exUROLOGY 56 (1), 2000

TABLE II. Prevalence of comorbidities in 50 consecutive patients with primary necrotizing fasciitis of the male genitalia Comorbidities Diabetes Cardiovascular Pulmonary History of alcoholism Compensated renal insufficiency Psychiatric illness Severe immunocompromise

n

%

Mortality

25 24 13 13 12

50 48 26 26 24

12 8 8 39 17

12 6

24 12

17 83

change unable to attain a predicted arterial oxygen pressure greater than 1000 mm Hg were excluded. HBO treatment consisted of two to three sessions of 90 minutes at 2.5 to 3 atm absolute within the first 24 hours, followed by re-evaluation of the patient and an individualized number of further courses of two sessions every 24 hours at 2 atm absolute until evidence of clinical stability. Some patients with underlying diabetes or vascular disease received additional treatment to accelerate granulation. The median number of HBO sessions was 6 (range 1 to 35). Of the 50 consecutive patients, 6 patients (12.0%) were found to be severely immunocompromised because of leukemia (n ⫽ 3) or systemic steroid treatment (n ⫽ 3) for lupus erythematosus, panhypopituitarism, or autoimmune hepatitis. These patients were already critically ill before the development of Fournier’s gangrene. Given their distinctly different presentation, clinical course, and dismal outcome, they were analyzed separately. Statistical analysis was performed using a commercial statistical software package. Mortality was defined as disease-related death during the hospital stay, and survival was measured in days. The correlation of prognostic variables (Table II) and patient survival were studied by univariate analysis using the Kaplan-Meier algorithm.10 Groups were compared with the log-rank test. Statistically significant variables were entered into a multivariate regression analysis according to the Cox hazard model.11 The distribution of the BSA was analyzed using the independent samples t test. P values are reported as the result of two-tailed testing, and P values less than 0.05 were considered statistically significant.

RESULTS The mean age of the 50 patients was 56.3 years (range 28 to 79). Comorbidities were prevalent, with diabetes mellitus the most common (Table II). The median number of surgical debridements performed was two and ranged from one to six. At the time of admission to our institution, 28 patients (63.7%) had undergone initial surgical debridement at an outside institution (outside hospital stay 2.9 ⫾ 0.46 days), with a median delay of 4 days after the initial symptoms presented. The auxiliary surgical measures performed consisted of urinary (n ⫽ 11, 25.0%) and fecal diversion (n ⫽ 18, 41.0%) (Table I). Of 44 patients, 38 (86.4%) subsequently received HBO treatment. The paUROLOGY 56 (1), 2000

tients’ morbidity resulted in a median stay in the surgical intensive care unit of 4.5 days and a prolonged median hospital stay at our institution of 26 days (range 1 to 78). Despite extensive therapeutic efforts, the mortality rate was high at 11.4% (5 of 44) among the immunocompetent patients eligible for outcome analysis. Thirty-nine patients (88.7%) had an identifiable colorectal (45.5%) or genitourinary (43.2%) source of infection. Most patients (n ⫽ 36, 81.8%) had positive intraoperative wound cultures. Blood cultures were positive in 8 patients (18.2%) (Table I). The extent of infection was small (BSA 0.5% to 3.0%) in 24 patients (54.5%), medium (BSA 3.5% to 5.0%) in 9 patients (20.5%), and large (BSA greater than 5.0%) in 11 patients (25.0%) (Table I). The mean involved BSA was calculated as 4.2% ⫾ 0.62%. All 24 patients with a small involved BSA of 3% or less survived. The patients with a medium and large involved BSA experienced a mortality rate of 11.1% (n ⫽ 9) and 36.4% (n ⫽ 11), respectively (Table I). Most cases of necrotizing fasciitis were classified as superficial (n ⫽ 31, 70.5%). There was a positive correlation between the extent and depth of involvement: 8 (72.7%) of 11 patients with a large extent had a deep necrotizing infection. Screening of the outcome variables by univariate analysis identified three statistically significant predictors of survival. These were the extent of involved BSA (small, medium, or large) (P ⫽ 0.0262), the depth of the necrotizing infection (P ⫽ 0.0107), and treatment with HBO (P ⫽ 0.0115). None of the other variables impacted significantly on survival (Table I). In the subsequent multivariate analysis, only the extent of the infection was identified as a significant, independent predictor of outcome (P ⫽ 0.0234) in the presence of other covariables. There was a trend toward an improved outcome in HBO-treated patients, but it did not reach statistical significance (P ⫽ 0.0677) (Table I). Six patients (12.0%) were severely immunocompromised and critically ill before developing Fournier’s gangrene and were analyzed separately. These patients were younger, with a mean age of 48.2 years versus 57.2 years in the immunocompetent group. In all 6 immunocompromised patients, the blood cultures were positive; however, only 8 (19.5%) of the other 44 patients had positive blood cultures. A large extent of disease was more frequent and occurred in 5 (83.3%) of the 6 immunocompromised patients compared with 11 (25.0%) of the 44 remaining patients. Finally, 83.3% (n ⫽ 5) of the 6 immunocompromised patients died compared with 11.4% (n ⫽ 5) of the 44 remaining patients. 33

TABLE III. Characterization of patients with Fournier’s gangrene in recent series (1989–2000) Series Pizzorno et al.,12 1997 Benizri et al.,14 1996 Baskin et al.,201990 Laor et al.,15 1995 Pai et al.,8 1996 Korhonen et al.,5 1998 Hejase et al.,13 1996 Clayton et al.,21 1990 Current series, 2000

Period

Patients (n)

ETOH (%)

DM (%)

GU CR (%) (%)

1990–1996 1978–1991 1975–1988 1978–1992 1984–1993 1971–1996 1993–1995 1976–1986 1984–1998

11 24 29 30 30 33 38 57 50

27 — 52 — — 12 66 51 26

73 29 41 30 30 21 66 32 50

— 21 48 — 0 — 32 46 38

— 50 20 — 20 48 — 33 40

Mortality (%)

Adverse Outcome Factors

0 25 21 43 20 9 3 16 20

No HBO Age, extent, positive BCx Age* FGSI*, age* WCx, Immunocomp., DM No HBO,* delay* Delay Age* Extent,† Immunocomp.

KEY: BCx ⫽ blood culture; CR ⫽ colorectal source of infection; DM ⫽ diabetes mellitus; ETOH ⫽ chronic alcohol abuse; FGSI ⫽ Fournier’s Gangrene Severity Index; GU ⫽ genitourinary source of infection; HBO ⫽ hyperbaric oxygen; Immunocomp. ⫽ severe immunocompromise; WCx ⫽ wound culture. * Significant by univariate analysis. † Significant by multivariate analysis.

COMMENT In our study of 50 consecutive patients treated for Fournier’s gangrene at a single institution, the local extent was a strong predictor of outcome both in the univariate and multivariate analyses. Notably, none of the 24 patients with an involved estimated surface area of 3.0% or less died. This observation concurs with the finding that patients with minimal disease extent of 1% or less have low mortality rates of 0%12 or 3%13 respectively. Although this association of outcome and extent of disease has previously been suggested,14 other investigators have been unable to demonstrate a statistically significant difference15 or have questioned the impact of the involved BSA on outcome.3 We found that the differences in BSA between survivors (3.6 ⫾ 0.47) and patients who died (9.2 ⫾ 1.75) was highly statistically significant (P ⫽ 0.0001). On the basis of the results in our series, it appears that patients with an involved BSA of 5% or greater are at particularly high risk of succumbing to the complications of Fournier’s gangrene. This observation concurs with findings of Palmer et al.3 that “patients with limited disease, i.e. less than 5% surface are likely to survive.” They go on to emphasize, “however, survival is not directly proportional to the surface area involved.” Although we consider the results of this study an important cornerstone in the growing body of evidence that the extent of disease may be the most important single prognostic variable, our results support the idea that the outcome of patients with Fournier’s gangrene is influenced by a number of other factors, including severe immunocompromise. The development of the Fournier’s Gangrene Severity Index Score (FGSI)15 constitutes an important contribution in the attempt to predict the outcome of individual patients on the basis of a number of variables rather than a single criteria. On the basis of the Acute Physiology and Chronic 34

Health Evaluation II (APACHE II) Symptom Score,16 the FGSI uses the degree of deviation from normal of important physiologic variables to generate a score that correlates with patient mortality.15 It remains for future studies to investigate whether the FGSI, as a general measure of physiologic derangement, is practical and accurate in assessing patients with Fournier’s gangrene. Given the impact of the disease extent on outcome, we believe that any severity of disease index for patients with Fournier’s gangrene should include this important prognostic variable. Although experimental data indicate some benefit from HBO to optimize tissue oxygenation,17 its overall impact in the treatment of Fournier’s gangrene is controversial. The indication for HBO in Fournier’s gangrene is of particular interest because of the high costs and limited availability of this form of adjunctive treatment. Pizzorno et al.12 recently reported a series of 11 patients who all underwent HBO therapy with no reported morbidity or mortality. This exceptionally low mortality rate, however, is likely attributable to the small mean involved BSA and that the series included only 1 patient with involvement beyond the external genitalia and/or perineum. In our series, the univariate analysis found a significant difference in the observed outcome of HBO-treated and nontreated patients (P ⫽ 0.0115). In the presence of other significant covariables in a multivariate analysis, the difference approached but did not meet statistical significance (P ⫽ 0.0677). It is possible that a larger series with greater power would have revealed a statistically significant difference. Given the limited treatment options currently available, we believe that patient treatment should be aggressive and should include HBO treatment. Our series of 50 patients included a subset of 6 severely immunocompromised patients with myeloproliferative disorders or who were undergoing treatment with systemic steroids. Few reports have so far UROLOGY 56 (1), 2000

highlighted the poor prognosis of Fournier’s gangrene occurring in the setting of the severely immunocompromised state of a patient with leukemia or steroid dependence.6,18,19 In our series, the course of these patients was characterized by early sepsis, positive blood cultures in all patients, and a statistically significant larger extent of infection. We believe these patients constituted a separate group with a particularly fulminant course of Fournier’s gangrene and an outcome predominantly determined by their underlying disease, which may in many cases be resistant to all therapeutic efforts. Potential limitations to the design and interpretation of this study deserve mention. First, this represents a retrospective study of patients treated during a long period. Second, a significant proportion of patients were initially treated at an outside institution and referred for further treatment. Patients with more advanced disease are therefore likely to have been over represented in this study, and pretreatment may have skewed the observed patient characteristics. Finally, the involved BSA at the time of maximum extent was determined retrospectively versus at the time of treatment. We do believe, however, that the available photo documentation and the operative reports allowed an accurate estimation of the true disease extent. Given the robust correlation of outcome to BSA in both univariate and multivariate analysis, any small variation is unlikely to have impacted significantly on the results of this study. CONCLUSIONS In the multivariate analysis of the prognostic variables, the extent of infection, as determined by the involved BSA, was a highly statistically significant, independent predictor of outcome. Routine documentation of the involved BSA may help identify high-risk patients and allow timely referral of these patients to specialized centers of care. On the basis of the trend of an improved outcome of HBO-treated patients in this and other reported series, HBO treatment of patients with Fournier’s gangrene, in particular those with an extension of 5% BSA or greater, appears indicated. Severely immunocompromised patients may constitute a subset of patients with a fulminant course and a dismal prognosis. REFERENCES 1. Stephens BJ, Lathrop JC, Rice WT, et al: Fournier’s gangrene: historic (1764 –1978) versus contemporary (1979 –1988) differences in etiology and clinical importance. Am Surg 59: 149 –154, 1993. 2. Fournier A: Gangrene foudroyante de la verge. Med Prat 589 –597, 1883. 3. Palmer LS, Winter HI, Tolia BM, et al: The limited impact of involved surface area and surgical debridement on survival in Fournier’s gangrene. Br J Urol 76: 208 –212, 1995. 4. Riseman JA, Zamboni WA, Curtis A, et al: Hyperbaric UROLOGY 56 (1), 2000

oxygen therapy for necrotizing fasciitis reduces mortality and the need for debridements. Surgery 108: 847– 850, 1990. 5. Korhonen K, Hirn M, and Niinikoski J: Hyperbaric oxygen in the treatment of Fournier’s gangrene. Eur J Surg 164: 251–255, 1998. 6. Berg A, Armitage JO, and Burns CP: Fournier’s gangrene complicating aggressive therapy for hematologic malignancy. Cancer 57: 2291–2294, 1986. 7. Hotter JT: Re: Fournier’s gangrene as the presenting sign of an undiagnosed human immunodeficiency virus infection. J Urol 155: 291–292, 1996. 8. Pai NB, Gerst PH, Yousuf AM, et al: Necrotizing fasciitis in immunocompromised patients. Contemp Surg 4: 12–14, 1996. 9. Pruitt BA, Goodwin CW, and Pruitt SK: Burns, in Sabiston DC (Ed): Textbook of Surgery, 15th ed. Philadelphia, WB Saunders, 1997, pp 221–252. 10. Kaplan EL, and Meier P: Nonparametric estimation from incomplete observations. J Am Stat Assoc 53: 457– 481, 1958. 11. Cox DR: Regression models and life tables. J R Stat Soc 34: 187–220, 1972. 12. Pizzorno R, Bonini F, Donelli A, et al: Hyperbaric oxygen therapy in the treatment of Fournier’s disease in 11 male patients. J Urol 158: 837– 840, 1997. 13. Hejase MJ, Simonin JE, Bihrle R, et al: Genital Fournier’s gangrene: experience with 38 patients. Urology 47: 734 – 739, 1996. 14. Benizri E, Fabiani P, Migliori G, et al: Gangrene of the perineum. Urology 47: 935–939, 1996. 15. Laor E, Palmer LS, Tolia BM, et al: Outcome prediction in patients with Fournier’s gangrene. J Urol 154: 89 –92, 1995. 16. Knaus WA, Draper EA, Wagner DP, et al: APACHE II: a severity of disease classification system. Crit Care Med 13: 818 – 829, 1985. 17. Grim PS, Gottlieb LJ, Boddie A, et al: Hyperbaric oxygen therapy. JAMA 263: 2216 –2220, 1990. 18. Wolach MD, MacDermott JP, Stone AR, et al: Treatment and complications of Fournier’s gangrene. Br J Urol 64: 310 –314, 1989. 19. Heurkens AH, Peters WG, van den Broek PJ, et al: Fournier’s gangrene or fulminant necrotizing fasciitis of the scrotum and penis as a complication of granulocytopenia in a patient with acute myelogenous leukaemia (AML). Neth J Med 32: 235–239, 1988. 20. Baskin LS, Carroll PR, Cattolica EV, et al: Necrotising soft tissue infections of the perineum and genitalia: bacteriology, treatment and risk assessment. Br J Urol 65: 524 –529, 1990. 21. Clayton MD, Fowler JEJ, Sharifi R, et al: Causes, presentation and survival of fifty-seven patients with necrotizing fasciitis of the male genitalia. Surg Gynecol Obstet 170: 49 – 55, 1990.

EDITORIAL COMMENT The authors have studied the outcomes of patients with necrotizing fasciitis of the male genitalia in a large, retrospective series accumulated during a 15-year period at a single institution. HBO was a distinctive treatment used in 86.4% of the patients, but it predicted a statistically significant improved outcome only with univariate analysis. Multivariate analysis, however, showed that only the extent of the involvement of the BSA was a statistically significant predictor of outcome. Within their own commentary, the authors refer to a series in which there was no mortality when HBO was used.1 The authors attributed this to the low BSA not to the HBO. Within their current series, however, the criteria for using HBO was “clinical evidence of a necrotizing mixed aerobic/ anaerobic infection” independent of the BSA involved. Seemingly, all patients diagnosed with Fournier’s gangrene were treated with HBO unless pulmonary contraindications ex35