Percutaneous Isolated Hepatic Perfusion for the Treatment of Unresectable Liver Malignancies
- 1.5k Downloads
Liver malignancies are a major burden of disease worldwide. The long-term prognosis for patients with unresectable tumors remains poor, despite advances in systemic chemotherapy, targeted agents, and minimally invasive therapies such as ablation, chemoembolization, and radioembolization. Thus, the demand for new and better treatments for malignant liver tumors remains high. Surgical isolated hepatic perfusion (IHP) has been shown to be effective in patients with various hepatic malignancies, but is complex, associated with high complication rates and not repeatable. Percutaneous isolated liver perfusion (PHP) is a novel minimally invasive, repeatable, and safer alternative to IHP. PHP is rapidly gaining interest and the number of procedures performed in Europe now exceeds 200. This review discusses the indications, technique and patient management of PHP and provides an overview of the available data.
KeywordsInterventional oncology Liver/hepatic Percutaneous hepatic perfusion Melphalan
The liver is frequently affected by cancer. Primary liver cancer is the sixth most common cancer in the world and the third cause of cancer-related death . The liver is also a predilection site for metastases from various malignancies [1, 2]. Surgery or ablation offers the best chance of a cure in most liver malignancies, but this is often not feasible due to the extend or location of the disease. Liver malignancies have a dominant or exclusive vascular supply from the hepatic artery, whereas 70–80 % of the supply of the non-tumorous liver parenchyma is derived from the portal vein [3, 4]. This difference in perfusion is utilized in liver-directed therapies, such as trans-arterial (chemo-) embolization or radioembolization. The unique hepatic anatomy also allows vascular isolation of the liver to deliver high doses of cytotoxic agents with minimal systemic toxicity. Isolated hepatic perfusion (IHP) is a complex surgical technique that involves clamping of the inferior vena cava (IVC) and portal vein (PV), ligation of IVC tributaries and arterial hepatico-enteric anastomoses with subsequent infusion of a high dose of chemotherapy into the proper hepatic artery [5, 6, 7, 8, 9, 10]. Promising results have been obtained with IHP in treating liver tumors from different histology. Response rates of 37–52 % have been reported for metastatic ocular melanoma patients [11, 12, 13, 14]. In patients with liver metastases from colorectal carcinoma, response rates of 50–60 % have been obtained [15, 16, 17]. Despite the good response rates, the complexity and duration (up to 9 h) of the procedure have prevented for IHP to gain wide acceptance . Furthermore, IHP is generally not repeatable and is associated with high morbidity and mortality rates [16, 18, 19, 20].
Percutaneous hepatic perfusion (PHP) is a novel alternative to IHP that enables vascular isolation and perfusion of the liver with the use of endovascular techniques . The minimal invasiveness as well as the repeatability of PHP offers an important advantage over IHP. This review will describe this highly innovative technique and provide an update of current literature on PHP.
Summary of conducted studies on percutaneous hepatic perfusion
No. pts (no. PHPs)
Type of hepatic malignancy (n)
Type of study
R % (n)
Hughes et al. 
93 (max 6 per pt)
Melanoma (ocular 83; cutaneous 10)
Response (primary: hPFS)
27.3 % (vs 4.1 % in control)
7.0 months (vs 1.6 in controls)
10.6 months (vs 10.0 in controls)Ω
Neutropenia (85.7 %), thrombocytopenia (80.0 %), anemia (62.9 %), self-limiting hyperbilirubinemia (14.3 %), cardiac toxicity (12.9 %), cerebral ischemia (1.2 %), death 3.2 %
Vogl et al. 
Melanoma (ocular 8; cutaneous 3), Gastric/Breast/CA (1)
Response and toxicity
Pancytopenia; death (7.1 %; retroperitoneal hemorrhage)
Fitzpatrick et al. 
Melanoma (ocular 4; cutaneous 1)
Feasibility and toxicity
Transient mild hypothermia and metabolic acidosis
Fukumoto et al. 
HCC, BCLC intermediate (27) or advanced (41)
Mitomycin C and/or doxorubicin
Resection + PHP
Leukopenia (44.1 %), serum AST gr. 3/4 (77.9 %), hair loss (72 %), gastroduodenal ulcer (4.4 %)
Forster et al. 
Melanoma (ocular 5; cutaneous 3, unknown 1). Sarcoma (1)
Response and toxicity
Bone marrow suppression; mild elevation serum troponin (70 %)
Pingpank et al. 
Acute transaminitis gr 3/4 (22 %); neutropenia 47 %, thrombocytopenia 29 %, anemia (15 %). Death 0.04 % (cholangitis)
Miao et al. 
Melanoma (ocular 12, cutaneous 4), NET (12), CRC (7), HCC (5), RCC (4), AdrC/Breast/CA (2), Ewing (1)
Hemodynamics and metabolic changes
Transient hypotension and metabolic acidosis; nausea/vomiting (10 %)
Melanoma (ocular 10, cutaneous 3), CRC (2), Hepatobiliary (3), NET (4), RCC (2), AdrC/Breast/Sarcoma/PA (1)
MTD, toxicity, pharmacokinetics
Neutropenia gr 3/4 (73.6 %); thrombocytopenia gr 3/4 (36.8 %); anemia (21.1 %)e
Ku et al. 
resection + PHP
1 and 5 years OS: 86 and 47 %
Leukopenia (45.5 %), hair loss (63.6 %)
Savier et al. 
IHP + PHP
Neutropenia grade 3/4 (50 %)
Ku et al. 
HCC, TNM III (1) or IV-A (27)
Response and survival
Chemical hepatitis (71 %), leukopenia (54 %), hair loss (43 %), thrombocytopenia (18 %), hemolysis/hematuria (57 %), gastroduodenal ulcer (7 %), death 8 % due to pancreatitis (4 %) and HAT (4 %)
Ku et al. 
HCC (11), CRC (1), Breast CA (1), Melanoma (1)
Hemodynamics, pharmacology, toxicity
Chemical hepatitis (75 %), leukopenia (43.7 %), alopecia (37.5 %), thrombocytopenia (25 %), hemolysis/hematuria (50 %)
Ku et al. 
HCC, unresectable (15)
Hemodynamics, pharmacology, toxicity, response
12 months for responders (vs 5 for non-responders)
Chemical hepatitis (71 %), leukopenia (67 %), alopecia (33 %), thrombocytopenia (40 %), hemolysis/hematuria (87 %), gastroduodenal ulcer (14 %), death 13.3 % due to pancreatitis (7 %) and HAT (7 %)
Ravikumar et al. 
HCC (5), CRC (8), Melanoma (2), Sarcoma (4), Adrenal/Pancreatic/SCLC/CA (n = 1)
Hematologic, primarily leukopenia/neutropenia; transient hypotension (78.5 %)
Hepatic Vascular Mapping
At present, only one PHP system is commercially available (Chemosaturation Hepatic Delivery System, Delcath Systems Inc, New York, USA), and therefore, some of the techniques described are specific to this system. In Japan, another double-balloon catheter (4L/2B, Fuji System Co. Ltd, Tokyo, Japan) is currently used in clinical studies.
Once correct positioning of the two balloons is confirmed, a stepwise approach is used to start filtration of blood by the two cartridges. A centrifugal pump is used to achieve a flow rate between 0.40 and 0.75 L/min. The maximal flow rate should not exceed 0.8 L/min and pre-pump pressures should not exceed −250 mmHg to avoid the catheter to collapse or kink. The hemofiltration filters are brought online one by one, by removing the clamps. Once the cartridges are completely filled with blood, the bypass line is closed. When the hemofiltration circuit is running sufficiently and hemodynamic stability is achieved (see below), intra-arterial infusion of chemotherapeutic drugs may be started using a pump injector and a flow rate of 0.4 mL/s. Before and during the infusion, hepatic angiograms are obtained to ensure that hepatic blood flow is not compromised. If the angiograms show arterial spasms, this may be treated with nitroglycerine boluses of 100–200 micrograms. After the infusion, extracorporeal filtration is continued for 30 min (‘washout period’) to allow clearance of chemotherapeutics from the liver. At the end of the procedure, the effects of heparin are reversed by administration of protamine sulfate on a 1:1 basis (1 mg of protamine sulfate to antagonize 1 mg of heparine). The vascular sheaths are left in place until coagulation is sufficiently corrected, although a vascular closure device may be placed immediately after the procedure to achieve hemostasis at the arterial puncture site. The duration of the procedure is generally 3–4 h.
Anesthesiology and Perfusionist Support
PHP is associated with hemodynamic and metabolic changes that require monitoring and management by an experienced anesthesiologist. In early studies, the conduct of PHP under local anesthesia and sedation has been described, but nowadays procedures are generally performed under general anesthesia [28, 29]. PHP results in significant decreases in mean arterial and central venous pressures and increases in heart rate compared to baseline [29, 30]. Decreases in blood pressures generally occur at two stages: upon occlusion of the IVC and when blood flow is diverted through the filters. Inflation of the balloons of the double-balloon catheter results in a decrease in central venous return and right atrium pre-load. This first drop in blood pressure is corrected by administration of fluids and norepinephrine and/or phenylephrine to maintain a mean arterial pressure above 60 mmHg. A second drop in blood pressure may be attributed to the depletion of sympathomimetics by the activated carbon filters. Infusion rates as high as 0.2–1.5 μg/kg/min for norepinephrine and 0.4–3.0 μg/kg/min for phenylephrine are generally required during the perfusion period as 67–95 % of the sympathomimetics are cleared from the blood by the filters . A decrease in patient’s body temperature is also commonly encountered during PHP and is a result of the blood flowing through the non-heated extracorporeal circuit. In general, the hypothermia is not severe and may be reduced by using an air-warming system .
The majority of studies have used melphalan chloride as the chemotherapeutic agent of choice, for it has pharmacological properties to make it suitable for PHP. It can easily be administered intra-arterially, has limited liver toxicity, a high hepatic extraction rate, a very short half-life, and an immediate effect on tumor cells [31, 32]. The currently available filtration system (Delcath hemofiltration cartridges) is specific for melphalan chloride.
Melphalan chloride is an alkylating agent of the nitrogen mustard group. Its binding to deoxyribonucleic acid (DNA) can result in cross-linking between bases on complementary strands leading to double-stranded DNA breaks and eventually cell death [8, 10, 33, 34, 35, 36, 37, 38]. In a phase I dose-escalating study with percutaneous administration, the maximum tolerated dose (MTD) of melphalan chloride was 3.0 mg/kg body weight . The maximum total dose is generally limited to 220 mg. Because of the short life of melphalan chloride, the drug should be prepared in the pharmacy just prior to administration.
Several studies have used doxorubicin as the chemotherapeutic agent, mainly in patients with hepatocellular carcinoma (HCC) [29, 40, 41, 42, 43]. Doxorubicin has some disadvantages over melphalan chloride as an agent for PHP. Firstly, it has a first-pass hepatic extraction fraction that is only around 60 % . Secondly, doxorubicin is associated with considerable liver toxicity. Studies on PHP with doxorubicin have reported chemical hepatitis rates of >70 % [29, 40, 41, 42, 43]. The chemical hepatitis was generally mild to moderate and self-limiting. Thirdly, with the currently available filters, the doxorubicin infusion time, and consequently the administrated dose, is limited as prolonged infusion may lead to increased systemic exposure. In a study by Ku et al., a mean doxorubicin extraction rate of 91 % was found, but the filtration rated dropped to 55 % at 20 min . Nevertheless, very promising results have been obtained with doxorubicin in patients with advanced HCC with PHP as either the primary treatment or as an adjunct to surgery (see results section). The hemofiltration cartridges (DHP-1; Kuraray Co., Ltd., Osaka, Japan) used in this study differ from those in studies with melphalan chloride.
Patients are monitored in a medium or intensive care unit (ICU) 12–24 h after the procedure and are generally discharged after 2–3 days. This compares favorably to IHP for which admission to the ICU is generally several days and a mean hospital stay has been reported of 10–29 days . Anemia, neutropenia, and thrombocytopenia may be seen early after the procedure and may (in part) reflect dilution as a result of peri-procedural fluid administration. Transfusion with fresh-frozen plasma, packed red blood cells, or platelets may occasionally be needed. PHP is associated with transient metabolic acidosis, but infrequently to such a degree that correction with sodium bicarbonate is required .
Table 1 provides an overview of the type and frequency of common and serious complications as reported in the literature. The most common adverse effects result from bone marrow suppression leading to neutropenia, thrombocytopenia, and/or anemia. Mild-to-moderate bone marrow suppression is seen in half to three-quarters of patients. The nadir of cytopenia is generally 10–14 days after PHP. It is generally recommended to administer granulocyte colony-stimulating factor analogues (pegfilgastrim) within 48 h after PHP to anticipate bone marrow depression. Symptomatic anemia and severe thrombocytopenia (<20,000/mm3) may require transfusions. Regular blood tests in the first 2 weeks after PHP are recommended.
Complications related to multiple vascular accesses and vessel catheterization might occur. Patients are at an increased risk of puncture site bleeding as high doses of heparin are administrated during the procedure to prevent clot formation in the extracorporeal circuit. Bleeding other than from the puncture site is uncommon, but may have severe consequences [29, 30, 46]. The hypotension associated with PHP may potentially result in complications such as organ ischemia. In general, the hypotension is of short duration and responds well to administration of fluids and sympathomimetics.
The reported mortality rate of PHP is 0–13.3 % (reference Table 1). Most of the published deaths occurred in early studies at the beginning of the learning curve and with a system that was different from the kit that is currently used. The reported IHP-related death rate is much higher than that of PHP: 5–27 % .
Leakage to the Systemic Circulation
Bone marrow depression is a result of leakage of melphalan chloride to the systemic circulation. Increased systemic exposure to melphalan chloride may be a result of incomplete filtration of the chemotherapeutic agent by the hemofiltration filters. In a phase I dose-escalating study, pharmacological blood samples were obtained during 74 procedures in 28 patients with unresectable hepatic malignancies . Perfusions were performed with Hemosorba drug filtration cartridges (Asahi Medical Co, Tokyo, Japan) for which the filter extraction percentages ranged from 58.2 to 94.7 %, with a mean of 77 %. A second-generation filter system is available since 2012, and this filter was reported to have an efficiency rate of 99 % in preclinical studies . Initial experiences with the second-generation filter seem to indicate that the degree of bone marrow depression with this filter is lower compared with that associated with previous filter systems .
There may be causes of melphalan chloride leakage other than through the filter system. The fact that, even with the second-generation filter, mild-to-moderate bone marrow depression is not infrequently seen seems to suggest that leakage other than through the hemofiltration system indeed occurs . One potential cause of leakage may be insufficient sealing of the balloon at the atriocaval junction with consequent leakage alongside the balloon. Furthermore, leakage to the systemic circulation could also be a result of the presence of collateral pathways between the IVC and azygos, hemiazygos, accessory hemiazygos, thoracolumbar, and/or diaphragmatic veins. Small interconnecting veins between the aforementioned structures are not uncommon and may cause the melphalan chloride to bypass the extracorporeal filter system. Another possible mechanism may be the uptake of melphalan chloride by the hepatobiliary system and storage until the balloons are deflated, after which melphalan chloride is released systemically. However, IHP is associated with lower rates of leakage of chemotherapeutic drugs than PHP . Furthermore, the half-life of melphalan chloride is very short. It therefore seems less plausible that post-procedural release of chemotherapeutics by the liver is the cause of systemic toxicity.
In IHP, leakage can be monitored by injection of human serum albumin (HSA) or erythrocytes labelled with iodine-31 or technetium-99 [45, 49, 50]. A closed, recirculating system is used in IHP, and detection of labelled HSA or erythrocytes in the systemic circulation is an indication of leakage. Unfortunately, this method cannot be applied in PHP as the perfusion circuit is not a closed system, and the activated carbon filters allow passage of both HSA and erythrocytes. Leakage can be quantified by measurement of systemic drug levels during PHP, but this does not provide real-time information as laboratory tests to determine melphalan plasma levels are rather complex and time-consuming.
Results of PHP to Date
In 2005, Pingpank et al.  published results of a phase I dose escalation study on PHP with melphalan chloride in 28 patients with primary and metastatic hepatic disease, establishing a MTD of 3 mg/kg. Response and survival rates were not primary endpoints, but were reported. In the 10 patients with metastases from ocular melanoma, an objective response rate (ORR) of 50 % was observed: two complete responses (CR) and three partial responses (PR). In the total study group, six PRs were documented (21.4 %). The duration of CR was 10 and 12 months, and duration of PR included two patients with ongoing responses at 9 and 11 months.
In a retrospective study by Forster et al., including 10 patients with hepatic metastases from ocular melanoma (n = 5), cutaneous melanoma (n = 3), melanoma from unknown origin (n = 1), or sarcoma (n = 1), nine patients (90 %) had stable disease or PR on follow-up imaging . The median percent decrease in hepatic tumor volume was 48.6 % for patients with ocular melanoma compared to 33.3 % for the entire cohort. At a median follow-up of 11.5 months (range 4–55 months), median hepatic-free survival was 240 days and median overall survival from the time of first PHP was 8.7 months.
A retrospective two-center study reported the results of PHP in 14 patients treated with 18 PHP procedures . The majority of patients (n = 11; 78.5 %) had liver metastases from melanoma [ocular (n = 8) or cutaneous (n = 3)]. A 50 % ORR was reported (one CR in a patient with cholangiocarcinoma and six PRs in patients with metastases from melanoma) and 38 % patients had stable disease.
Recently, the results were published of a multi-center, randomized controlled study comparing PHP with best alternative care (BAC) in patients with hepatic metastases from melanoma . The study included 93 patients with unresectable hepatic metastases from either ocular (n = 83) or cutaneous (n = 10) melanoma. Patients with limited extrahepatic disease were allowed to enter the study, although most patients (59.1 %) had metastases confined to the liver. Patients in the PHP arm (n = 44) underwent a maximum of six isolated liver perfusions with melphalan at 4–8 weekly intervals. Patients in the control group (n = 49) received best alternative care (BAC) with the majority of patients (81.6 %) receiving active treatment such a systemic chemotherapy, chemoembolization, radioembolization, and surgery. A statistically significant improvement in hepatic progression-free survival (hPFS) and overall progression-free survival (oPFS) was demonstrated in patients treated with PHP compared to BAC. The hPFS and oPFS were 7.0 and 5.4 months respectively for the PHP group compared to 1.6 and 1.6 months respectively for the BAC group (p < 0.0001). No statistically significant difference in overall survival (OS) was found between the PHP and BAC group (10.6 and 10.0 months respectively), but this was confounded by a large proportion of the patients in the BAC group (57.1 %) crossing over to receive PHP after progression of disease.
A Japanese group has published several studies on PHP in patients with HCC using a different double-balloon catheter and hemofiltration system (see above). Some overlap between the patient groups in the different studies exists [42, 54, 55, 56]. In a prospective study, 28 patients with advanced HCC (TNM III or IV-A) underwent an average of 1.4 PHP with doxorubicin. The ORR was 63 % and the OS was 16 months. The 1-, 3-, and 5-year survival rates were 67.5, 39.7, and 39.7 %, respectively . In a recent publication, the results were reported of combined reductive surgery and PHP with mitomycin C and/or doxorubicin in 68 patients with intermediate- or advanced-stage HCC . An ORR of 70.6 % was achieved with a median OS of 25 months.
Summary of ongoing prospective studies on percutaneous hepatic perfusion
Type of hepatic malignancy
Phase II, single center
PHP with melphalan. 2 cycles
ORR, post-PHP resectability safety, OS, HPFS, PFS, QoL
Phase III, multicenter
PHP with melphalan. Max 6 cycles versus BAC (dacarbazine, TACE, ipilimumab or pembrolizumab)
OS, PFS, ORR, HPFS, hepatic ORR, QoL
Launching fourth quarter 2015
Phase II, single center
PHP with melphalan. 2 cycles
ORR, post-PHP resectability safety, OS, HPFS, PFS, QoL
Phase II, multicenter
HCC or ICC
PHP with melphalan. 2 Cycles
ORR, safety, PFS
Phase II, multicenter
PHP with melphalan. 3 cycles, followed by sorafenib
Adverse events, ORR, PFS, pharmacokinetics, QoL
PHP followed by sorafenib
PFS, OS, safety
In conclusion, PHP is a novel, minimally invasive, and repeatable alternative to IHP. Phase I studies have demonstrated PHP to be feasible and safe. A recently published randomized controlled trial has shown improved control of liver disease compared to standard available therapy in patients with hepatic metastases from (ocular) melanoma. Further phase II and III studies are needed to define the role of PHP in the clinical management of patients with different hepatic malignancies.
The authors thank Gerrit Kracht for producing the figures.
Compliance with Ethical Standards
Conflict of interest
The Leiden University Medical Center has received financial support from Delcath Systems Inc. for conducting studies on PHP. All authors declare that they have no conflicts of interest.
This article does not contain any studies with human participants or animals performed by any of the authors.
Does not apply.
- 24.Miyayama S, Yamashiro M, Okuda M, Yoshie Y, Sugimori N, Igarashi S, et al. Usefulness of cone-beam computed tomography during ultraselective transcatheter arterial chemoembolization for small hepatocellular carcinomas that cannot be demonstrated on angiography. Cardiovasc Intervent Radiol. 2009;32(2):255–64.CrossRefPubMedGoogle Scholar
- 26.Takayasu K, Muramatsu Y, Maeda T, Iwata R, Furukawa H, Muramatsu Y, et al. Targeted transarterial oily chemoembolization for small foci of hepatocellular carcinoma using a unified helical CT and angiography system: analysis of factors affecting local recurrence and survival rates. AJR Am J Roentgenol. 2001;176(3):681–8.CrossRefPubMedGoogle Scholar
- 32.Vahrmeijer AL, van Dierendonck JH, Keizer HJ, Beijnen JH, Tollenaar RA, Pijl ME, et al. Increased local cytostatic drug exposure by isolated hepatic perfusion: a phase I clinical and pharmacologic evaluation of treatment with high dose melphalan in patients with colorectal cancer confined to the liver. Br J Cancer. 2000;82(9):1539–46.CrossRefPubMedPubMedCentralGoogle Scholar
- 39.Pingpank JF, Libutti SK, Chang R, Wood BJ, Neeman Z, Kam AW, et al. Phase I study of hepatic arterial melphalan infusion and hepatic venous hemofiltration using percutaneously placed catheters in patients with unresectable hepatic malignancies. J Clin Oncol. 2005;23(15):3465–74.CrossRefPubMedPubMedCentralGoogle Scholar
- 42.Fukumoto T, Tominaga M, Kido M, Takebe A, Tanaka M, Kuramitsu K, et al. Long-term outcomes and prognostic factors with reductive hepatectomy and sequential percutaneous isolated hepatic perfusion for multiple bilobar hepatocellular carcinoma. Ann Surg Oncol. 2014;21(3):971–8.CrossRefPubMedGoogle Scholar
- 47.Moeslein FM, McAndrew EG, Appling WM, Hryniewich NE, Jarvis KD, Markos SM, et al. Evaluation of Delcath Systems’ Generation 2 (GEN 2) melphalan hemofiltration system in a porcine model of percutaneous hepatic perfusion. Cardiovasc Intervent Radiol. 2014;37(3):763–9.CrossRefPubMedPubMedCentralGoogle Scholar
- 53.Hughes MS, Zager J, Faries M, Alexander HR, Royal RE, Wood B, et al. Results of a randomized controlled multicenter phase III trial of percutaneous hepatic perfusion compared with best available care for patients with melanoma liver metastases. Ann Surg Oncol. 2015. doi:10.1245/s10434-015-4968-3.
- 57.Fitzpatrick M, Richard Alexander H, Deshpande SP, Martz DG Jr, McCormick B, Grigore AM. Use of partial venovenous cardiopulmonary bypass in percutaneous hepatic perfusion for patients with diffuse, isolated liver metastases: a case series. J Cardiothorac Vasc Anesth. 2014;28(3):647–51.CrossRefPubMedGoogle Scholar
- 58.Pingpank JF, Royal RE, Kammula US, Kam AW, Wood BJ, Libutti SK, et al. Chemo-saturation with percutaneous hepatic perfusion (CS:PHP) using Melphalan for unresectable neuroendocrine tumor liver metastases (MNET). Abstract CIRSE 2011; FP2109.Google Scholar
- 60.Kawai S, Tani M, Okamura J, Ogawa M, Ohashi Y, Monden M, et al. Prospective and randomized trial of lipiodol-transcatheter arterial chemoembolization for treatment of hepatocellular carcinoma: a comparison of epirubicin and doxorubicin (second cooperative study). The Cooperative Study Group for Liver Cancer Treatment of Japan. Semin Oncol. 1997;24(2 Suppl 6):S6-38–45.Google Scholar
Open AccessThis article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made.