Atraumatic Splenic Rupture in Legionella pneumophila Pneumonia

A previously fit 46-year-old male handyman presented to a rural hospital with a cough, fever, and epigastric pain without peritonism. The patient was admitted medically with symptoms and radiological appearances consistent with atypical community-acquired pneumonia. During the first 48 hours of admission, he suffered a significant haemodynamic deterioration and was transferred to the intensive care unit (ICU) for vasoactive support. Following stabilisation, urgent abdominal CT imaging demonstrated splenic rupture with haematoma in the absence of historical trauma. Emergency splenectomy was performed; the histopathological examination was unremarkable. Investigations for the presenting complaint confirmed Legionella pneumophila serotype 1 pneumonia by urinary antigen testing. The patient was extubated on postoperative day 2 and stepped down from ICU to complete a 14-day course of azithromycin. Atraumatic splenic rupture is a rarely described clinical entity. The process can be subdivided into pathological and nonpathological (spontaneous) cases. Pathological atraumatic splenic rupture may occur in the context of wide-ranging aetiologies, including bacterial pneumonia; however, the association with Legionella pneumophila serotype 1 is exceptional, with this representing the eighth case in the medical literature.


Introduction
Community-acquired pneumonia (CAP) is a leading cause of hospital admissions, mortality, and health-related cost worldwide [1]. While Streptococcus pneumoniae is the leading cause of pneumonia worldwide, approximately 15-20% of CAPs are secondary to atypical organisms [2]. Atypical pneumonias are named such due to their relative resistance to beta-lactam therapy (resulting from a lack of a bacterial cell wall), their intracellular life cycle, their inability to be cultured or gram stained using traditional methods, their diverse symptomatology, and their propensity for extrapulmonary disease. Of these atypical pathogens, Legionella pneumonia is the most common [3,4].
Clinically, Legionella infection may present with Pontiac fever or pneumonia. Pontiac fever is an acute self-limiting febrile illness with a paucity of respiratory tract signs; [5][6][7] Legionella pneumonia however presents 2-10 days postexposure with dry cough on the background of a prodrome of headache, myalgia, arthralgia, diarrhoea, and relative bradycardia. Chest X-ray fndings may range from isolated pulmonary infltrates to well-circumscribed lesions and cavitation [8,9]. Clinical diferentiation from other causative pathogens of CAP may be suggested by the presence of haemoptysis, transaminitis, or hyponatraemia due to syndrome of inappropriate ADH release (SIADH) [8].
Idiopathic ASR is defned according to the Orlof and Peskin criteria [11]. Tese comprise an absence of trauma and perisplenic adhesions and normal macroscopic/histological examinations of the abdominal viscera including the spleen. More recently, a ffth criterion of negative serology for relevant viral infections has been proposed [12].
Management of ASR can vary from conservative management, to interventional radiological techniques such as arterial embolization, and in more catastrophic cases, emergent laparotomy and total splenectomy. Te decision regarding embolization Vs surgical approaches may be guided by logistical factors, such as local interventional radiology availability, transport times to referral centres, and the degree of patient haemodynamic instability [12]. In this case report, we present a case of ASR in a 46-year-old noncomorbid gentleman diagnosed with Legionella pneumophila pneumonia. Tis is the eighth recorded case in the literature.

Case Report
A 46-year-old Caucasian male with no past medical history, regular medications, or illicit drug use presented to a rural emergency department with fve days history of rigors, night sweats, and unrelenting cough despite a course of oral antibiotics provided by their primary care physician.
Te patient reported a 15-pack year smoking history and avian contact in the weeks preceding his illness. Te patient was a handyman by occupation with duties including water system maintenance, but had no recent foreign travel.
On presentation, the patient was febrile (40.1°C), diaphoretic, and tachycardic (sinus rhythm, 140 beats per minute), but normotensive (140/70 mmHg). He was tachypnoeic (20-30 bpm), saturating 94% on room air, with a productive cough and streaks of haemoptysis. Plain chest radiograph performed in the emergency department revealed right perihilar consolidation and air bronchograms, consistent with clinically auscultated crepitations over the right lower lung feld. Te diagnosis of community-acquired pneumonia (CAP) was made ( Figure 1). Te patient reported some mild abdominal discomfort, with palpation positive for LUQ and epigastric tenderness, without peritonism, guarding, nor organomegaly. Tere were no stigmata of infective endocarditis nor exanthem. His SMART-COP score was 2 on admission with a CURB65 score of 1.
After crystalloid fuid resuscitation, benzylpenicillin and doxycycline were administered per local guidance, providing coverage for typical and atypical causes of CAP. Te patient was referred and admitted under the general medical team for ongoing treatment. Respiratory virus testing, sputum culture, and blood cultures were requested along with urinary pneumococcal and Legionella urinary antigens. Serological testing for atypical pathogens including Mycoplasma pneumoniae and Chlamydia psittaci was also sought in view of avian exposure, and the presence of extrapulmonary symptoms. Infectious mononucleosis was ruled out with nonreactive viral capsid antigen IgM and negative monospot with cytomegalovirus and HIV serology yielding negative results.
24 hours later, the patient was attended by the medical emergency response team for a new episode of hypotension (BP 86/51 mmHg) and worsening tachypnoea during which high-fow nasal oxygen was required. Due to progressive deterioration, an intensive care unit (ICU) opinion was sought, and he was transferred to the intensive care unit (ICU) for haemodynamic support presumed secondary to severe respiratory sepsis. Repeat haemoglobin measurement on admission to the ICU was decreased now to 94 g/L (haematocrit of 0.27) from 133 g/L one day prior. In the absence of melaena, injury, peritonism, or suggestion of haemolysis, this decrease was initially attributed to haemodilution. Lactate per arterial blood gas (ABG) was slightly increased at 2.1 mmol/L. Central venous access was obtained and vasopressors commenced; however, there was a rapid escalation in noradrenaline and subsequent vasopressin infusions, and thus stress dose hydrocortisone was supplemented for the presumption of severe vasoplegic septic shock. Due to this deterioration, antimicrobial therapy was escalated to piperacillin/tazobactam 4.5 g TDS and IV azithromycin 500 mg OD as per local guidelines for severe CAP. Following initial stabilisation, four hours into intensive care admission the patient decompensated suddenly with a frankly peritonitic abdomen. Tey required endotracheal intubation for humanitarian reasons and worsening type 1 respiratory failure. Urgent computed tomography pulmonary angiogram (CTPA) and contrast abdominalpelvis CT were performed posthaemodynamic resuscitation, demonstrating multilobar pulmonary consolidation, with large volume perihepatic fuid, and heterogeneous splenic enhancement consistent with laceration and haemoperitoneum ( Figure 2). In the absence of a trauma history, abrupt and progressive haemodynamic instability and radiological fndings a diagnosis of ASR was made.
Arterial blood gas analysis performed postreturn from the imaging department demonstrated severe raised anion gap metabolic acidosis and anaemia: pH 7.03, lactate 13.5 mmol/L, BE −19.5 mEq/L, and Hb 63 g/L. Te massive transfusion protocol was activated, while arrangements were made for emergent transport to the operating theatre. At laparotomy, intraoperative fndings confrmed capsular rupture of the spleen, and 4 L of altered blood product was retrieved from the abdomen, following which the surgical team proceeded to perform a total splenectomy. Macroscopic and histopathological examination revealed a spleen of normal architecture, weighing 137 g, with no evidence of malignant infltration.
Postoperatively, the metabolic acidosis rapidly improved, with a minimal transfusion requirement of 2 further packed red cell units in the following 24 hours. On the second postoperative day, the patient was successfully extubated and liberated from all haemodynamic supports. On day 3, urinary antigen for Legionella pneumophila returned positive, and thus, azithromycin was continued to complete a 14-day course. Te patient recovered uneventfully postsplenectomy and received the requisite vaccinations and ongoing antimicrobial prophylaxis as per national guidance and was discharged home.

Investigations
Initial lab tests (Table 1) revealed leucocytosis with neutrophil predominance, mild lymphopenia, and no peripheral eosinophilia. Infammatory markers were elevated, but venous lactate was normal at 1.5 mmol/L. Te patient had serum sodium at the lower limit of normal, mild acute kidney injury, and transaminitis. A rapid severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) antigen test was negative. In the setting of his occupation, the possibility of Legionnaire' pneumonia was raised.

Discussion
On admission, our patient was determined to have low CURB65 and SMART-COP scores which are prognostic in the setting of typical CAP. Te CURB65 score is a validated scoring system that helps determine the appropriate disposition of patients (inpatient treatment vs. outpatient) and mortality risk. Te score of 1 recorded in this patient was supportive of a low-risk course, appropriate for outpatient management [13]. Te SMART-COP score is a more sensitive scoring system in predicting the need for intensive care management, despite which our patient was designated again as low risk. Importantly, neither scoring systems are validated in atypical CAP [13,14] nor inform of the risks of rare complications such as atraumatic splenic rupture. Some 20-80% of patients with Legionella pneumonia progress to requiring intensive care support, and clinical deterioration can occur shortly after admission, as was the case in our patient. Mortality with timely investigation and management is reportedly 5-30%, but may be as high as 80% in immunosuppressed individuals experiencing treatment delays [15,16].
Splenic rupture is a surgical emergency and may present with abdominal pain with radiation to the left shoulder tip (Kehr's sign) [17]. Tis can be quickly accompanied by haemodynamic instability due to life-threatening intraabdominal haemorrhage. Unfortunately, clinical signs may be difcult to elicit in sedated, intubated patients in intensive care. In a series of 8 cases of atraumatic splenic ruptures of varying aetiologies, all patients had splenomegaly present (>200 g in weight or 11 × 7 × 5 cm in size) with common signs and symptoms of nausea, bowel obstruction, delirium, and peritonism (n � 4); however, Kehr's sign was rarely present (n � 2) [18]. In the setting of hypotension of unclear cause, with an associated reduction in haemoglobin, surgical causes including ASR should be considered [19]. Tis may also present atypically with thoracic back pain and right upper quadrant pain presumably due to haemoperitoneum [20]. While there was no prior abdominal imaging in our patient to exclude preexisting splenic pathology, the normal histopathology at the operation makes this unlikely. Our patient exhibited mild left upper quadrant and epigastric pain on presentation, but without attendant peritonism or guarding, and this was not considered sufcient to prompt targeted radiological investigation in the emergency department or during medical  Case Reports in Infectious Diseases admission. Te later development of Kehr's sign following admission to intensive care was coincident with substantial haemodynamic deterioration. Tis clinical emergency required resuscitation and stabilisation measures prior to performance of CT imaging being practicable or safe. While there was potentially a diagnostic delay, there is a lack of widespread appreciation of this rare complication of atypical pneumonia, and in the absence of trauma or convincing abdominal signs there was limited clinical suspicion for splenic rupture. Furthermore, at presentation, there may have simply been pain related to splenic capsular stretch from splenomegaly rather than rupture. It is possible that abdominal imaging on presentation may not have yielded fndings which would have altered the management course and that rupture may have only been provoked by coughing in the setting of pneumonia or Valsalva eforts later in the admission. Indeed, in patients with ASR, approximately a third of cases present with shock, yet the diagnosis is often made postmortem [21]. It would seem prudent as awareness of this entity grows, in the presence of left upper quadrant abdominal pain and pneumonia, that abdominal ultrasound be performed to facilitate early identifcation of patients at risk of ASR [22]. In all seven cases of ASR complicating Legionella pneumonia identifed in a literature search, Legionella pneumophila serotype 1 was the causative species (Table 2) [23][24][25][26][27][28][29]. While this may be due to specifc virulence factors, it may simply refect testing bias due to the global adoption of the urinary Legionella antigen test as the main method for diagnosing Legionella infection. Te urinary antigen test is an ELISA, with reported sensitivity of 94.6% and specifcity of 100% for diagnosing Legionella pneumophila serotype 1 but is of limited utility with respect to other pathogenic Legionella species such as serotype 2 and longbeachae [30,31]. Of the 7 previous cases we identifed (Table 2), 4 used the urinary antigen test with the remainder utilising direct immunofuorescence which is both technically challenging and of limited sensitivity [32]. In one instance, polymerase chain reaction (PCR) was also used to make the diagnosis. Tis is a highly specialised investigation and is not routinely available in most centres; however, PCR can be used on a wide range of clinical samples (e.g., blood, bronchoalveolar lavage fuid, and sputum) and can detect all relevant Legionella species and serogroups and thus may become a gold standard test in the future [33]. Interestingly, in all previously identifed cases of ASR complicating Legionella infection, patients were male with an average age of 55 years (range 42-70 years). Te majority (71%) had no predisposing comorbidities for Legionella infection, in keeping with our report. Te remaining two cases were immunocompromised by diabetes mellitus (n � 1) and methotrexate use (n � 1), of whom one case died [26,29].
Te interval between the onset of illness and ASR varied but was often within the frst week of illness (median 6 days). However, in one instance, splenic rupture occurred 3 weeks after apparently successful treatment for Legionella disease, highlighting the possible need to counsel patients regarding this complication as we do patients with infectious mononucleosis [29]. Overall survival of these identifed cases was 75% (n � 6/8) with all cases being managed surgically and receiving appropriate antibiotic regimes. In those who survived, there was a paucity of follow-up data, limiting inferences about long-term outcomes. In the two cases that died, diagnostic delay, uncontrolled bleeding, and severe comorbidities including immunocompromise and cardiomyopathy were likely contributory to poor outcomes [23,26]. In one instance, coinfection with H1N1 infuenza was noted alongside vascular thrombosis which may have precipitated splenic injury by means of portal hypertension [28]. Table 1: Admission blood tests and SMART-COP score.

Parameter Result
White cell count (WCC) 13.5 × 10 9 /L (3.6-11 × 10 9 /L) 89.8% neutrophils, 0.6 × 10 9 /L lymphocyte count C-reactive protein ( While it remains unclear which risk factors are most relevant to ASR pathogenesis in the setting of Legionella pneumonia, in a guinea pig model there was a clear relationship between Legionella inoculum size, disease severity, and the occurrence of death [34]. Potentially, the same holds true for the immunologic insult leading to ASR. ASR has been described in multiple settings in which malignancy, viral infection, infammatory and iatrogenic causes (e.g., antiplatelets and anticoagulants) predominate [11,35]. Other at risk groups include women during late pregnancy (3 rd trimester) or those with ectopic pregnancies in whom this diferential should be considered given the raised intra-abdominal pressure from the gravid uterus [36]. More recently, ASR has been described in both children and adults with SARS-CoV-2 of severity ranging from asymptomatic to requiring invasive mechanical ventilation [37,38]. Te pathophysiology is thought to be from COVID-19-induced lymphocyte necrosis, follicle depletion, splenic blood vessel congestion, and vascular proliferation with resultant haemorrhage [39]. As such, SARS-CoV-2 testing should be included as part of the viral serological testing in patients presenting with ASR.
Of the four cases of Legionella-associated ASR where histopathological examination of the spleen was performed, fndings ranged from scanty polymonorphonuclear infltrate [23], splenomegaly [25], congested infamed red pulp (splenitis) [27], and difuse splenic coagulation [28]. While in our patient histological analysis was determined normal, it is unclear if interobserver variability or insufcient sampling may explain these fndings. Te underlying mechanism of ASR in Legionella likely involves downstream aberrant immunological responses with splenic hyperplasia, coagulopathy from septicaemia, and venous congestion from associated cold agglutinin formation. In this setting, any sudden increase in portal pressure from coughing or abdominal palpation may precipitate rupture [40].
Given the characteristics of Legionella, appropriate antimicrobial agents must be able to penetrate the lung parenchyma and achieve high intracellular concentrations. Antibiotic classes which have inherent activity against Legionella include the macrolides, rifamycins, tetracyclines, and quinolones [8]. Levofoxacin and Azithromycin are frstline therapies favoured by the BTS (British Toracic Society) and IDSA (Infectious Diseases Society of America), respectively, given their excellent pharmacokinetics and inherent bactericidal activity. When comparing quinolone and macrolide efcacy in Legionella infection, a recent highly powered systematic review found there to be no signifcant diference in mortality (6.88% versus 7.43% and p � 0.66) or overall treatment success between the two groups [41]. Indeed, in most clinical, outcomes there was no signifcant diference between the two classes of antibiotics although fuoroquinolones were associated with reduced complication rates including respiratory failure and subsequent Clostridioides difcile infection which is a major consideration given its associated morbidity and mortality [41]. While resistance to these agents in Legionella infection has rarely been reported, this may become an issue in the future given the empiric use of Legionella active therapy in the initial management of CAP globally and is worthy of future attention. While the recommended duration of therapy varies, it should be typically provided for 5-10 days but this may be extended up to 21 days in those who are immunosuppressed or with severe disease, as evidenced in our case by the need for both intensive care admission and splenectomy [32].

Conclusion
Tis report adds to the small number of cases of L. pneumophila 1 pneumonia complicated by ASR and highlights the importance of Hickams' Dictum and multidisciplinary consultations. In patients with Legionella pneumonia and left upper quadrant pain, ASR should be considered and investigated. Using traditional criteria for operative intervention, patients with ASR are more likely to require splenectomy. With timely diagnosis and prompt surgical intervention, a favourable outcome can be achieved.

Conflicts of Interest
Te authors declare that they have no conficts of interest. Case Reports in Infectious Diseases 5