Conflicting study data about Merck KGaA's (MRK.XE) key cancer drug proved a rollercoaster for shares of the German pharmaceutical company Wednesday.

Shares fell Wednesday afternoon after an independent study showed Erbitux failed to improve survival rates in colon cancer patients compared with standard chemotherapy.

The results seemed to contradict a Merck study published earlier in the day - which sent shares up - that showed colon cancer patients lived 3.5 months longer on Erbitux compared with chemotherapy.

Analysts say confusion about the data weighed on the shares. At 1331 GMT, the shares were down EUR2.75, or 4%, at EUR66.02, while the DAX was up 0.2%.

The independent COIN study showed patients who received Erbitux lived a median rate of 17 months, compared with 17.9 months for those on chemotherapy alone, and wasn't statistically significant.

Earlier Wednesday, Merck said data from a trial called CRYSTAL showed colon cancer patients lived 3.5 months longer - 23.5 months compared with 20 months - when Erbitux was added to their chemotherapy treatment.

The drug is already approved for colon cancer in the U.S. and Europe so the data have no impact on where the drug is sold. However, the data could give competitors such as Amgen Inc.'s (AMGN) Vectibix a marketing edge depending on how the market interprets the results.

"The difference between the CRYSTAL and COIN studies lies in that one was commissioned by the company; the other was independent," said Sal. Oppenheim analyst Peter Duellmann.

However, Duellmann said the CRYSTAL data have greater meaning and higher clinical relevance.

The results were presented at the European Society for Medical Oncology congress in Berlin.

Erbitux, a monoclonal antibody whose active ingredient is called cetuximab, is already approved as a primary treatment in combination with chemotherapy for colon cancer. Analysts consider the drug a potential blockbuster, with peak annual sales of at least EUR1 billion. It is marketed in the U.S. by Eli Lilly & Co. (LLY) and Bristol-Myers Squibb Co. (BMY).

"The COIN study is disappointing for us because we couldn't reach the primary endpoint of showing an advantage in overall survival," Wolfgang Wein, executive vice president of oncology for Merck's pharmaceutical division, Merck Serono, told Dow Jones Newswires in an interview Wednesday.

Wein attributed the difference between the studies to a weakened patient population in the COIN study.

The independent COIN study included about 2,400 colon cancer patients and had three arms: a group of patients who continued receiving oxaliplatin-based chemotherapy; a group who received Erbitux and oxaliplatin therapy; and a third group who received oxaliplatin-based chemotherapy intermittently.

Wein said patients in the Erbitux arm were given significantly less chemotherapy.

"Additionally we've seen that significantly less chemotherapy was given in the Erbitux arm and also in the second-line therapy so that the Erbitux effects apparently washed off," he said.

Wein said the company must analyze the therapy strategy for the study for older, weaker patients.

"We must understand exactly what occurred in the study, and what it means for our general treatment strategy when significantly less chemotherapy was given in the Erbitux arm and furthermore for patients in very bad shape."

-By Heide Oberhauser-Aslan and Allison Connolly, Frankfurt Bureau; +49 69 29725500; heide.oberhauser@dowjones.com