We identified 30 patients with PHL within our institutional cohort and 192 patients from comprehensive literature review ( Table 1 ). Of our institutional PHL cohort, the predominant histology types included 15 (50%) patients with DLBCL, 5 (17%) MZL, 3 (10%) FL, 4 (13%) CLL/SLL, 2 (7%) T-cell lymphoma, and 1 (3%) had low-grade B-cell lymphoma, NOS.

Overall demographics, disease characteristics and response are noted in Table 2 . Median follow-up was 6 years. Sixteen patients were male (53%), 25 (84%) were Caucasian, 10 (33%) had a history of smoking, 3 (10%) had prior hepatitis C infection and 2 (7%) had prior hepatitis B infection. PHL was mostly identified incidentally (9 patients; 30%), with 8 (27%) presenting with abdominal pain and 6 (20%) presenting with weight loss. Other presenting symptoms included fatigue (4 patients; 13%), jaundice (3 patients; 10%), diarrhea (3 patients; 10%), and nausea/vomiting (3 patients; 10%). At the time of diagnosis, median LDH was 261 (range 127-2116), median hemoglobin was 12.9 (range 8.4-15.9), and median platelet count was 231 (range 78-432).

Most patients had an ECOG performance status of 0 (25 patients; 83%), and Lugano Stage I (25 patients; 83%), with 3 patients (10%) with Stage III, and 2 (7%) with Stage IV disease. Six (20%) patients had bulky disease measuring ≥7.5cm and 3 (10%) presented with nodal involvement. One (3%) patient had concurrent splenic disease. Systemic therapy was the most common treatment modality overall (20 patients; 67%) with surgery, radiation and observation utilized in 4 patients (13%) each. Seventeen (57%) of patients were alive at the time of data collection, with 8 (27%) deceased and 5 (17%) lost to follow-up. Cause of death for the 8 patients included 5 disease-related deaths and 3 unrelated to PHL. The 5 disease-related deaths included deaths from infection while on immunosuppression after transplant, 1 from infection while receiving systemic therapy, 1 from liver failure, and 1 from widespread progression of disease. The 3 patients with deaths unrelated to PHL included 2 deaths from complications of other unrelated medical comorbidities and 1 patient with no evidence of disease 12 years after last cancer treatment. On univariable analysis, only age (p=0.009; 95% CI 0.030, 0.190) and ECOG score (p=0.048; 95% CI 0.616, 1.000) were found to be significantly associated with overall survival. No variables were found to be significant on multivariable analysis.

In the paired literature review, 96 patients (50%) received systemic therapy, 69 (36%) underwent surgery, 10 (5%) received radiation and 11 (6%) underwent observation ( Table 2 ). Seventeen (57%) of patients were alive and 5 (17%) lost to follow-up within our institutional cohort, compared with 97 alive patients (51%) and 3 (2%) lost to follow-up in the paired literature review.

We performed a subgroup analysis of DLBCL patients (n=15; Table 3 ), including 7 (44%) with Germinal Center B-cell lymphoma (GCB), 2 (13%) with Activated B-cell Lymphoma (ABC), and 6 (38%) with NOS. Patients were predominantly male (63%) and white (81%). Patients had overall good performance status with 73% of patients with ECOG 0 and 20% of patients with ECOG 1. Lymphomas were generally early stage (73% with Lugano Stage I) including 38% of patients with bulky disease. Almost all patients received systemic therapy at some point (93%) and one patient was observed (7%). Two patients underwent surgery (13%) and none received radiotherapy at any point. ECOG score and Lugano stage were associated with OS on UVA but did not remain significant on MVA.

Six DLBCL patients with available targeted sequencing data were identified at our institution, with 4 mutations mutated more than once: TP53 (50%), MEF2B (33%), P2RY8 (33%), and IRF8 (33%). Pathway mapping of genetic mutations was notable for alterations in cell cycle pathway genes TP53 and ATM (17%). No other pathways had more than one genetic alteration.

Subgroup analysis of the indolent lymphoma group included a total of 9 patients total, including 5 (56%) with MZL, 3 with FL, and 1 with low-grade BCL, NOS ( Table 4 ). Among them, 4 were male, 8 were of white ethnicity, and the mean age was 67 years (range 42-77). All patients were ECOG 0, 89% had Lugano Stage I disease and no patients had bulky disease. Initial treatment modalities were varied including 2 patients each (22%) who received radiation therapy, systemic therapy, surgery or observation ( Figure 1 ). Both patients who underwent radiotherapy monotherapy received a low-dose regimen of 4Gy in 2 fractions with complete response and no evidence of disease at 15 and 18 months respectively. Three patients were observed (33%). No patients received combined chemotherapy and radiation as first-line treatment, however one patient received combined chemoradiotherapy as a second-line treatment for lung recurrence. Complete response was seen in 6 patients, progression in 2 patients, and 1 patient was lost to follow-up. Both patients who progressed were initially treated with systemic therapy. Of those, one patient was treated with 6 cycles of R-CHOP with CR, followed by transformation to DLBCL 22 months later for which they received systemic therapy. The other patient was initially treated with rituximab with short-interval recurrence 4 months later for which they received 2 additional systemic therapy regimens as well as radiotherapy to the lungs bilaterally ( Figure 1 ).

Median survival in our cohort was 6 years, compared with 2 years among the literature review cohort.

To build upon our institutional experience, we conducted a literature review to contextualize our experience within the broader literature. Existing literature consists largely of small patient cohorts or individual patients with heterogenous treatment approaches based on primary histology. Below we describe the basic demographics, clinical presentations, treatment approaches and outcomes, when available, categorized by lymphoma histology.

DLBCL is the predominant histology observed among primary hepatic lymphomas (6, 10). We found 75 cases of primary hepatic DLBCL in 64 studies, with a predominance among male gender (male to female ratio: 1.27), and an mean age of 60.6 years (range 19-82).

Some patients had concomitant cancers including esophageal stromal tumor (11), tumor forming pancreatitis (12) and HCC with gastric adenocarcinoma (13). One patient had a synchronous DLBCL lymphoma in the colon (14). Viral infections such as HCV (13, 15–24), HBV (11, 15, 18, 25–28), HIV (29) and EBV (28, 30, 31) were the most common predisposing factors. Other factors such as auto-immune and hepatic disease were also observed in some cases, including alcoholic cirrhosis (32), primary biliary cirrhosis (33), recurrent hepatic IPT (12), Sjogrens syndrome (33–35), sarcoidosis (36), and polymyositis (30, 37). Five patients received methotrexate for treatment of polyarthritis (27, 28, 36, 38, 39). One patient was noted to have a history of IV drug use (19).

The main clinical presentation varied among patients. Many reported GI symptoms such as right upper quadrant pain (14, 18, 19, 26, 28, 29, 34, 36–38, 40–56), abdominal discomfort (57–60), jaundice (40, 41, 47, 57, 61–63), melena (14), and left upper quadrant pain (64).

Nineteen patients presented with fever (11, 21, 30, 36–40, 46, 48, 50, 51, 53, 63, 65–68), 11 with weight loss (11, 14, 18, 19, 39, 43, 46, 56, 59, 62, 69), one with night sweats (34), and only 4 cases had all B symptoms (29, 50, 53, 60). Other systemic symptoms such as loss of appetite (70) and altered mental status (66, 67) were present. The hepatic mass was incidentally found on imaging for routine follow-up of a chronic disease in 10 cases (13, 17, 23–25, 27, 31, 32, 50, 71), and, in one case, during a workup of a pancreatic mass (12).

On physical examination, one study described a right upper quadrant mass bulge in 7 patients (15) and 2 cases were found to have hepatomegaly (61, 65). One patient presented with generalized edema (72). In 2 studies, patients were diagnosed after presenting with abnormal laboratory results (elevated LFTs (67) and cytopenia (30)).

In 2013, Kashimura M. et al. reported 2 cases where the diagnosis was made on autopsy. Both cases presented initially with fever and elevated LFTs with normal imaging: One was treated with corticosteroids for hypercalcemia, and the other with antibiotics for cholecystitis and peritonitis (67).

First-line treatment strategies were heterogeneous but were mainly chemotherapy-based. Few patients underwent partial hepatectomy alone (11, 15, 43, 52, 73). Methotrexate was discontinued in patients who were receiving it as the only intervention (28, 30), or as adjunct to other treatment such as surgery (27), R-THP-COP (rituximab, pirarubicin, cyclophosphamide, vincristine, and prednisolone) (38), R-CHOP (39), and one month of prednisolone (36).

CHOP regimen was used alone (14, 22), or preceded by 1 cycle of BACOP (bleomycin, driamycin, cyclophosphamide, vincristine, and prednisone) (17) or CEOP (cyclophosphamide, etoposide, vincristine, and prednisone) (23), or in combination with partial hepatectomy (20, 46, 53), or radiation therapy (one patient received 30.6Gy/Fr (65), another received 5x1.8Gy per week with 18Mv IMRT for a total dose of 31Gy (31)). The most commonly used chemotherapy regimen was R-CHOP (13, 15, 25–27, 30, 33, 35, 46, 48, 51, 58, 60, 63, 65). In some cases, it was given prior to resection, without{sp}69 {/sp}or with radiofrequency ablation (44) or radiation therapy (19) (1.5Gy in 4 fractions using a 2 field 3D conformal technique with 10MV photon beam therapy on day 1, 2, 9 and 12). In other instances, it was combined with partial hepatectomy (18, 48, 49), or as part of other regimens. One patient received R-CHOP and was maintained on cyclophosphamide alone for 1 year (58), while another patient received 1 cycle of R-CHOP followed by 5 cycles of R-MegaCHOP and intrathecal chemotherapy (56). Three patients were treated with R-THP-COP alone (55, 61), or with surgery (13). One patient underwent partial hepatectomy followed by COP therapy (60), another received rituximab with bendamustine (70), and one patient was treated with cyclophosphamide and methylprednisolone (59). Only one patient underwent autologous stem cell transplant, for which they received induction COP, R-CHOP then MCE (ranimustine, cyclophosphamide, etoposide) (71).

Six patients were not treated for several reasons: being clinically unstable (33, 68), lost to follow-up (41), patient refusal due to pregnancy (42, 72), and premature death (37).

Patients were followed for an average of 24 months (range 1-80, median: 22). Death occurred in three patients from aplasia and sepsis after R-CHOP (47, 50) or cyclophosphamide and methylprednisone (59). Partial regression was noted in one of the patients who received R-THP-COP (61), and in one of the methotrexate withdrawal group treated with corticoids alone, who experienced complete remission (CR) after R-CHOP regimen (36). An absence of response with tumor progression was reported in one patient treated with R-CHOP alone. However, the patient maintained complete remission for 3 years (25) after receiving second-line R-HyperCVAD (rituximab, cyclophosphamide, vincristine, doxorubicin, methotrexate, cytarabine and dexamethasone)/R-HD MTX ara-C regimen (high dose methotrexate and cytarabine) with radiation therapy (40Gy for 28 fractions). Tumor recurrence was noted in 3 cases. The first patient received R-CHOP and suffered multiple organ failure (45), the second one received autologous stem cell transplant followed by allogeneic bone marrow transplant second-line (71), and the third patient received rituximab and bendamustine and had lymph node involvement. This patient died after receiving R-miniCHOP (rituximab, cyclophosphamide, doxorubicin, vincristine and prednisolone) and 2 cycles of R-EPOCH (etoposide, prednisone, vincristine, cyclophosphamide, and doxorubicin) (70). Some articles did not provide information regarding outcomes (15, 18, 52, 64, 73). The remainder of the described cases were associated with CR ( Table 5 ).