http://www.turkgastro.org/text.php?id=428
This article states, "Of all tumors whose primary focus was determined, metastasis of pulmonary tumors was the most frequent."
That article is poorly written. If you really care about the answer to this (if I was you, I would just remember whichever source you trust more and move on, don't spend too much time on this) you can look at their data, which is so haphazard as to be nearly uninterpretable (at least, if you are looking for epidemiologic data like this)
I think they are trying to say something different there. Table two says that of the 704 total metastatic lesions, 443 were adenocarcinomas, 95 "malignant epithelial tumors" (whatever that means), 45 squamous cells, and 43 small cells. They only give the primary source of 194 of the adenocarcinomas, which means most were unknown primary. Table 3, which is entitled "Distribution of metastatic liver adenocarcinomas by their primary foci" has lung 4%, stomach 20%, colon 22%, breast 20%, ovaries 12%. You can parse this data all you want but basically it means that they didn't tell us the right stuff that you need to know to answer your question. So drawing any conclusions from this paper is not advised.
The WHO, who is basically the definitive source for this kind of thing, says the order of frequency is upper GI tract (stomach, gallbladder, pancreas; 44-78%); colon 56-58%, lung 42-43%, breast 52-53%, esophagus 32-33%, GU tract 24-38%. My interpretation (since if you add all those up you get well over 100%) is that these numbers refer to percentage of cases of each type of tumor that eventually metastasize to the liver. Actual incidence of tumor types can be found from the CDC or the WHO. In 2004, they say there were 145,000 colorectal tumors, and 196,000 lung tumors. Multiply by above percentage and that means there were 82000 lung cancers that metastasized to the liver and about about 82000 colon cancers that metastasized. Which one is more? Depends whose math you use. If you use the breast numbers (189,000 tumors) you get 98,000. So by that logic breast is more common. Except for the fact that it isn't in real life.
The refereces, by the way, that the WHO gives for this data, are all at least 20 years old. One of them is from 1977 and is incidence in Malmo, another is 1982, and the third and most respected is the textbook published by the Armed Forces Institute of Pathology, in which they use autopsy data from 1150 patients and state that lung is 24.8%, colon 15.7%, pancreas and breast 10% each, followed by others including 5% unknown.
All I would say is that in clinical practice, it seems as though colon is more common.
The other problem I have from all this data is that I highly doubt that these numbers are right anymore. Cancers are being caught earlier. Do 50% of breast cancers eventually metastasize to the liver? No way! It seems like more than 50% of breast cancers are cured these days by surgery from early detection. But I might be wrong there - old data is old data, remember.
What do you conclude from this? Stop worrying about it so much. If you get the question on Step I, remember it, and write them a letter of complaint in which you say that the data is conflicting and you want to know the real answer, or that they should throw out the question because there is no answer. Or you can just think like the question writer who probably assumes that colon is the most common (because clinically when you are faced with a liver mass and you are considering figuring out what it is and doing something about it, it's probably colon. If lung cancers spread to the liver, usually they have mets elsewhere and it is not a diagnostic dilemma) and just say colon.
Or go with the AFIP and say lung. Either way, just complain about the data because nothing has really been published in the past 20 years on this issue, and no doubt it has changed.
The most important thing to remember: Metastatic tumors are the most common type of liver tumor. Don't forget that one.